Facts

Dictation: facts and numbers

390 facts, each traced to the exact words of its source.

82 participants82 of the 96 volunteers in the 2010 habit study provided enough data for analysis.Phillippa Lally et al., European Journal of Social Psychology, 2010
96.3 percent of drafts with an error96.3 percent of the speech recognition drafts contained at least one error, against 58.1 percent after transcriptionist editing and 42.4 percent of signed notes.Li Zhou et al., JAMA Network Open, 2018
A 1996 study says that the performance of a speech recognition system degrades rapidly in the presence of ambient noise.Sang-mun Chi and Yung-Hwan Oh, International Conference on Spoken Language Processing (ICSLP) 1996, ISCA Archive, 1996
839 assessmentsA 2006 meta-analysis of distributed practice covered 839 assessments from 317 experiments in 184 articles.Nicholas J. Cepeda et al., Psychological Bulletin, 2006
A 2017 SAE paper concludes that AI% should be the preferred metric for relating cabin noise to speech recognition performance when the noise is steady.SAE International (Technical Paper 2017-01-1864), 2017
180 degrees off-axisA cardioid microphone is least sensitive to sound arriving from directly behind it, at 180 degrees off-axis.Shure Incorporated, Shure, Performance and Production insights, n.d.
180 words per minute (company claim)A court reporting company says a trained voice writer can sustain accuracy above 95 percent at speeds above 180 words per minute.Grace Johnston, Steno (court reporting company), Steno Brief, 2026
A homophone is a word that is pronounced the same as another word but differs in meaning or in spelling.Wikipedia contributors, Wikipedia, The Free Encyclopedia, n.d.
A May 1878 Chicago Tribune report of Edison's article said that phonograph dictation would let business people dispense with the clerk.Chicago Tribune; Phonozoic Text Archive, Document 059, 1878
A Philips foot pedal starts, pauses, rewinds and fast-forwards audio with a slight movement of the foot tip, leaving the hands free for typing.Philips Speech, 2026
A Radiology Today article advises radiologists that finding a quiet spot to dictate helps their concentration and produces a more accurate document.Beverly Brooks, Radiology Today, Vol. 9 No. 16, n.d.
1900 yearA review of Cita Stelzer's book Working with Winston says that from 1900 on, all of Churchill's writings and speeches were dictated to secretaries.Richard M. Langworth, Hillsdale College, Churchill Centre site, 2019
A stenomask is a handheld device that fits over the reporter's mouth, with a padded, soundproof enclosure and an internal microphone, which captures the voice for speech recognition and dampens the sound so others in the room cannot hear it.Grace Johnston, Steno (court reporting company), Steno Brief, 2026
A university help article says iOS and iPadOS 18 added on-device transcription of recordings.Bowdoin College IT knowledge base, n.d.
A university help article says that on iPhone you open a recording, choose Edit Recording, and tap the speech bubble icon to generate a transcript.Bowdoin College IT knowledge base, n.d.
141 yearsAbout 141 years passed between Edison's 1878 list of phonograph uses and Google's 2019 on-device speech recognizer.Calculated from other facts
6,200 openings per year, projectedAbout 6,200 openings for medical transcriptionists are projected each year, on average, over the decade.U.S. Bureau of Labor Statistics, 2025
2/3 of composition time spent planningAbout two-thirds of the time spent composing a letter was planning, whether the letter was written, dictated or spoken.John D. Gould and Stephen J. Boies, Science, 1978
According to Wikipedia, a voice writing system consists of a stenomask, an external sound digitizer and speech recognition software.Wikipedia, 2026
0.35 word error rateAcross five commercial ASR systems, the average word error rate was 0.35 for Black speakers and 0.19 for white speakers.Allison Koenecke et al., Proceedings of the National Academy of Sciences, 2020
0.3 percent of words in signed notesAfter a medical transcriptionist edited the drafts, the error rate fell to 0.4 percent, and to 0.3 percent in the notes physicians signed.Li Zhou et al., JAMA Network Open, 2018
1 hour median report turnaround (from 24)After speech recognition was introduced at a 150-bed community hospital, the median radiology report turnaround fell from 24 hours to 1 hour.Luciano M. Prevedello et al., Journal of the American College of Radiology, 2014
All verbal reporting procedures in the 2011 meta-analysis tended to increase the time needed to complete a task.Mark C. Fox et al., Psychological Bulletin, 2011
Apple advises that a headset microphone may help when Dictation is used in a noisy environment or a room with excessive echo.Apple Inc., Apple Support, macOS User Guide, n.d.
Apple advises turning on Command mode so that Voice Control only responds to commands and does not type them as text.Apple Support, Mac User Guide (macOS 27 version of the page), 2026
Apple gives composing messages and notes as examples of general text Dictation on a Mac, and says dictating in a search box is not part of that example.Apple Support, n.d.
Apple says Bluetooth has two modes, and the second is the one used for speaking through the microphone while also listening to audio.Apple, Apple Support, n.d.
Apple says Dictation is not available in all languages or regions, and that features may vary.Apple Support, n.d.
Apple says Dictation on a Mac also adds certain punctuation automatically as you dictate, without a spoken command.Apple Inc., 2026
Apple says Dictation on a Mac may need an internet connection, and that Keyboard settings show whether your dictation is processed on the device.Apple Inc., Apple Support, macOS User Guide, n.d.
Apple says Dictation on iPhone is processed on the device in many languages, and that no internet connection is required.Apple Inc., Apple Support, iPhone User Guide, n.d.
Apple says dictation sent to its servers is not stored unless you opt in to Improve Siri and Dictation.Apple, Apple, 2026
Apple says that if the iPhone is not detected by CarPlay, you should check whether CarPlay is restricted under Screen Time, in Content and Privacy Restrictions and Allowed Apps.Apple Support, 2026
Apple says that if the microphone does not work in an app, you should go to Settings, tap Microphone and make sure the app is enabled.Apple Inc., Apple Support, n.d.
Apple says that if you choose Press Fn Key Twice as the Dictation shortcut on a Mac, the matching Keyboard settings option changes automatically to Start Dictation (Press Fn Twice).Apple Support, n.d.
Apple says that in supported languages, Dictation on Mac adds certain punctuation automatically as you dictate.Apple Inc., Mac User Guide, 2026
Apple says that on iPhone you can use typing and Dictation together, because the keyboard stays open during Dictation so you can switch between touch and voice to enter text.Apple Inc., 2026
Apple says that once Voice Control is on, you can speak commands to navigate the desktop and apps, interact with what is on the screen, and dictate and edit text.Apple Support, Mac User Guide (macOS 27 version of the page), 2026
Apple says that when Bluetooth switches to its second mode, audio quality is reduced until the microphone is no longer in use.Apple, Apple Support, n.d.
Apple says that when you dictate text on a Mac, you can use commands to enter punctuation, capital letters and symbols, including symbols for math and currency.Apple Inc., 2026
Apple says the microphone source in Keyboard settings shows which device the Mac is currently using to listen for Dictation.Apple Support, n.d.
Apple says Voice Memos transcription on iPad is available on supported models in English and nine other languages.Apple Support (iPad User Guide), 2026
Apple says you can decide which apps are allowed to use the microphone on a Mac.Apple Inc., Apple Support, macOS User Guide, n.d.
Apple says you can keep your journal private, so that only you can access it, with Face ID, Touch ID or your passcode.Apple Support, iPhone User Guide, n.d.
Apple says your device shows in Keyboard settings whether dictated audio and transcripts are processed on the device or sent to Apple servers.Apple, Apple, 2026
Apple's CarPlay troubleshooting steps say to restart the iPhone and the car, and to make sure Siri is on, when CarPlay or Siri is not working.Apple Support, 2026
Apple's first suggested fix for reduced Bluetooth headphone audio on a Mac is to quit any apps that use the microphone of the headphones.Apple, Apple Support, n.d.
Apple's Mac Help says you can customize the key combination that starts and stops Dictation.Apple Support, n.d.
Apple's Pages guide says to start dictation in a Pages document by choosing Edit > Start Dictation, and that guided setup appears first if you have not used dictation before.Apple, Apple Support, Pages User Guide for Mac, n.d.
21.6 decibelsApplying the inverse-square rule to a microphone at 12 inches instead of 1 inch gives a level difference of about 21.6 dB, under ideal free-field conditions.Calculated from other facts
15 characters per reportAt the start, new users of speech recognition corrected 15 characters in a report of 433 characters.Tomi A. Kauppinen et al., Journal of Digital Imaging, 2013
90 percent accuracyAUDREY recognized the digits zero to nine with 90 percent accuracy, but only when its inventor spoke.Computer History Museum, n.d.
Azure AI Speech offers a removed option that drops all profanity from the recognition result.Microsoft Learn (Azure AI Speech), n.d.
Azure AI Speech's display text applies inverse text normalization, which converts spoken text to shorter forms, such as 200 for two hundred.Microsoft Learn (Azure AI Speech), n.d.
2.0 maximum weightAzure AI Speech's phrase list weight is set within a range of 0.0 to 2.0, where 0.0 disables the list and 2.0 is the maximum influence.Microsoft Learn (Azure AI Speech), n.d.
Azure AI Speech's profanity setting defaults to masked, which replaces profanity with asterisks.Microsoft Learn (Azure AI Speech), n.d.
Azure AI Speech's profanity setting has a raw option, which includes profanity in the recognition result, so the words are returned as recognized.Microsoft Learn (Azure AI Speech), n.d.
16 kHz sampling rateBauer et al. state that dictation and other large-vocabulary applications principally use a sampling rate of at least 16 kHz, especially in noisy environments.Patrick Bauer et al., EUSIPCO 2014, European Signal Processing Conference, 2014
Before continuous recognition, dictation products either had small vocabularies or made users pause between every word, which many found cumbersome.Susan Hura, Speech Technology Magazine, 1998
1952 yearBell Laboratories built AUDREY, its digit recognizer, in 1952.Computer History Museum, n.d.
1907 yearBy 1907, Henry James's dictation had become a confirmed habit, and his typist said its effect could be seen in his style.Theodora Bosanquet, Hogarth Press (Project Gutenberg edition), 1924
15 to 25 percent of the potential marketBy 1945, Dictaphone estimated that it had captured only 15 to 25 percent of its potential market.Allison Marsh, IEEE Spectrum, 2026
By default, Google Cloud Speech-to-Text leaves punctuation marks out of recognition results, so you must request automatic punctuation.Google Cloud, 2026
Charlotte Mason said the earliest writing practice for children of seven or eight should be transcription, not letter-writing or dictation.Charlotte M. Mason, Project Gutenberg, eBook 71087, 1886
10 minutesCharlotte Mason wrote that after the preparation of a dictation lesson, which takes ten minutes or less, there is rarely an error in spelling.Charlotte M. Mason, Project Gutenberg, eBook 71087, 1886
Charlotte Mason wrote that dictation, as commonly practised in schools in her time, was perhaps the most mischievous of the exercises children did at school.Charlotte M. Mason, Project Gutenberg, eBook 71087, 1886
Charlotte Mason wrote that every misspelt word leaves an image in the child's brain that the right spelling does not erase, so the common practice of correcting dictation errors makes bad spellers.Charlotte M. Mason, Project Gutenberg, eBook 71087, 1886
Charlotte Mason wrote that transcription should be an introduction to spelling, with the child looking at a word and then writing it from memory.Charlotte M. Mason, Project Gutenberg, eBook 71087, 1886
Chromebook dictation does not work if the language you speak does not match the device language.Google (Chromebook Help), 2026
20 hoursCiting an earlier study, Kraal and colleagues report that users who keep dictating for about 20 hours start to combine voice correction with keyboard and mouse correction.Ben Kraal et al., Proceedings of the 9th Australian International Conference on Speech Science and Technology, Melbourne, 2002, 2002
Clark and Fox Tree proposed that speakers use uh and um to announce a delay in speaking: a minor one they expect after uh, and a major one after um.Herbert H. Clark and Jean E. Fox Tree, Cognition, 2002
5 to 12 percentComposing by speech software at 8 to 15 words per minute runs at about 5 to 12 percent of normal speaking speed.Calculated from other facts
3.7 percent word error rateDecoding close-talking Wall Street Journal sentences recorded in a studio gave a word error rate of 3.7%, using DNN models trained on clean WSJ data.Mirco Ravanelli et al., arXiv (preprint, eess.AS), 2017
Dictaphone adopted magnetic tape, in the form of cassettes, in the 1960s.Allison Marsh, IEEE Spectrum, 2026
Dictaphone Corporation's 1997 annual report says the traditional desktop and portable dictation products used analog magnetic tape to store and replay voice.Dictaphone Corporation, U.S. Securities and Exchange Commission (EDGAR), 1997
Dictaphone dominated the market so thoroughly that its name became the generic term for dictation machines.Speech Technology magazine, n.d.
15 minutes per beltDictaphone introduced the plastic Dictabelt to the public in 1947, and each belt held about 15 minutes of recording.Allison Marsh, IEEE Spectrum, 2026
1947 yearDictaphone phased out wax cylinders in favor of the plastic Dictabelt in 1947.Speech Technology magazine, n.d.
695 US dollars at launchDragon NaturallySpeaking, released by Dragon Systems in 1997, cost $695 at launch and its price later fell to $349.Stephanie Schorow, Boston Herald; MIT Spoken Language Systems news archive, 1997
0.25 seconds of pause between wordsDragonDictate recognized discrete utterances, which the 1990 report defines as words with at least a quarter-second pause between them.Deseret News, Deseret News, 1990
9000 US dollars at launchDragonDictate, launched in 1990, cost $9,000 as a system of software and a speech recognition board.Deseret News, Deseret News, 1990
8 to 15 words per minuteEarlier reported composition speed using speech software was 8 to 15 words per minute, which the authors describe as much lower than normal speaking speed.Jinjuan Feng et al., Interacting with Computers, 2005
33 and 42 percent lexical density (speech and writing)Eggins (1994), as cited by Ishikawa (2015), found in a corpus analysis that lexical density was 33 percent in speech and 42 percent in writing.Shin'ichiro Ishikawa, Procedia - Social and Behavioral Sciences, 198 (2015) 202-210, 2015
4 percent decline, 2025 to 2035Employment of medical transcriptionists is projected to decline 4 percent from 2025 to 2035.U.S. Bureau of Labor Statistics, 2025
Fluid dictation in Windows voice typing corrects grammar, punctuation and filler words as you speak.Microsoft, 2026
1 to 3 inchesFor a handheld microphone, Nuance's Dragon guidance is to keep the microphone element 1 to 3 inches from the mouth, not directly in front of it.Nuance Communications, Nuance Communications, Dragon for Mac online help, n.d.
For high school students with learning disabilities, essays dictated to speech recognition were better than handwritten ones, and essays dictated to a human scribe were better still; students without learning disabilities showed no difference.Charles A. MacArthur and Albert R. Cavalier, Exceptional Children, 2004
94 studiesFox, Ericsson and Best's meta-analysis covered 94 studies that compared performance during concurrent verbalization with a silent condition.Mark C. Fox et al., Psychological Bulletin, 2011
From the earliest dictation machines, typists transcribed recordings on playback machines with foot pedals that controlled speed and let them reverse and repeat.Allison Marsh, IEEE Spectrum, 2026
100 words per minuteGoa's official scheme of examination for the Junior Stenographer post sets the shorthand skill test at 100 words per minute.Government of Goa, Government of Goa, 2021
2019 yearGoogle announced in 2019 an end-to-end, all-neural, on-device speech recognizer to power speech input in Gboard.Johan Schalkwyk, Google Research Blog, 2019
Google Cloud's Speech-to-Text API infers periods, commas and question marks from audio and adds them to the transcript when automatic punctuation is on.Google Cloud, 2026
Google says its voice typing and captions work in Google Docs and in the speaker notes and captions of Google Slides. Its help page for the feature does not list Google Sheets.Google (Docs Editors Help), 2026
Google says model adaptation by default has a relatively small effect on recognition, especially for one-word phrases.Google Cloud documentation, n.d.
3 browsersGoogle says voice typing in Google Docs works with the latest versions of Chrome, Edge and Safari, and that the web browser controls the speech-to-text service.Google (Docs Editors Help), 2026
Google's Docs help page, in its December 2020 archived copy, says voice commands are not available in Google Slides speaker notes, although voice typing can be used there.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
Google's Docs help page, in its December 2020 archived copy, says voice commands for editing and formatting a document are available only in English.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
Google's Docs help page, in its December 2020 archived copy, says voice typing works in a specific list of languages and accents, and gives that list on the page.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
Google's Docs help page, in its December 2020 archived copy, says you add punctuation in voice typing by saying specific phrases, and lists them.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
Google's Docs help page, in its December 2020 archived copy, says you can type and edit by speaking in Google Docs or in Google Slides speaker notes.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
Google's model adaptation boost feature increases the recognition model's bias toward some phrases more than others.Google Cloud documentation, n.d.
Ctrl+Shift+S keyboard shortcutGoogle's shortcut list for Docs gives Ctrl + Shift + S on Windows and Command + Shift + S on a Mac to start voice typing, labeled as available in Chrome browsers.Google (Docs Editors Help), 2026
1907 year (approximate)Graphophone dictating machines began to use the Dictaphone trademark around 1907, and in any case by 1908.The Office Museum, n.d.
Henry James's typist wrote that he found dictation an easier and more inspiring way to compose than writing by hand, even though he said he was too diffuse when dictating.Theodora Bosanquet, Hogarth Press (Project Gutenberg edition), 1924
Henry James's typist wrote that he said he was too diffuse when dictating, and that he considered the gain in expression more than compensated for any loss of concision.Theodora Bosanquet, Hogarth Press (Project Gutenberg edition), 1924
Henry James's typist wrote that he worked out the scenes of a book in detail before he began dictating the book itself.Theodora Bosanquet, Hogarth Press (Project Gutenberg edition), 1924
10 percent word errors (upper bound)High school students with and without learning disabilities learned to use speech recognition with total word errors under 10 percent.Charles A. MacArthur and Albert R. Cavalier, Exceptional Children, 2004
16 spoken wordsIBM's experimental Shoebox, designed in 1961, recognized 16 spoken words.IBM Archives, IBM MediaCenter, n.d.
IBM's Tangora, a research system meant to act as a voice-activated typewriter, had to be trained by each individual user, so it was speaker-dependent.B. H. Juang and Lawrence R. Rabiner, Elsevier Encyclopedia of Language and Linguistics (author copy, UC Santa Barbara), 2005
10 uses listed for the phonographIn 1878, Thomas Edison listed ten uses for the phonograph, and letter writing with dictation was at the top of the list.The Office Museum, n.d.
In 1983, the authors stated that speech recognition was not yet advanced enough to provide people with a reliable listening typewriter.John D. Gould et al., Communications of the ACM, 1983
In 2014, a Pennsylvania law firm had its lawyers record notes and thoughts on smartphones and upload the audio files to the firm's data center for transcription.BizTech Magazine, 2014
2015 yearIn 2015, a design firm worked with radiology staff at Texas Children's on concerns about noise control and lighting in a shared reading room.AuntMinnie.com, 2015
18.67 percent less task time with speech commandsIn a 1992 study, performance times for word-processing tasks were significantly faster when speech activated commands, with an average reduction of 18.67 percent in task time.Lewis Karl et al., Unpublished manuscript, University of Maryland (PDF hosted by Ben Shneiderman's site), 1992
In a 1999 CHI study, multimodal correction was faster than speech-only correction but slower than correction by typing for users with good typing skills.Bernhard Suhm et al., Proceedings of CHI 99, ACM Conference on Human Factors in Computing Systems, 1999
In a 2002 Virginia Tech study, postsecondary students with learning disabilities wrote higher-quality essays, at a faster rate, with continuous speech recognition than with a keyboard.Richard Conrad Snider, Virginia Tech (VTechWorks repository), 2002
In a 2006 review of verbal recall studies, the gap between study sessions that gave the best retention grew as the retention period grew.Nicholas J. Cepeda et al., Psychological Bulletin, 2006
42 voice-recognition MRI reports with errors (one radiologist)In a 2010 MRI audit of 50 voice-recognition reports and 50 transcriptionist reports, 42 and 30 finalised voice-recognition reports for the two radiologists contained errors, against 6 and 8 transcriptionist reports.Rodney H. Strahan and Michal Elisabeth Schneider-Kolsky, Journal of Medical Imaging and Radiation Oncology, 2010
96 volunteersIn a 2010 study, 96 volunteers each chose one behavior and repeated it daily in the same context for 12 weeks.Phillippa Lally et al., European Journal of Social Psychology, 2010
-0.03 correlation coefficient (r)In a 2011 meta-analysis, the effect of think-aloud, saying your thoughts while you do a task, on performance was statistically indistinguishable from zero.Mark C. Fox et al., Psychological Bulletin, 2011
11 characters per reportIn a 2013 hospital study, new users of speech recognition corrected 11 characters per report, on average.Tomi A. Kauppinen et al., Journal of Digital Imaging, 2013
In a 2015 laboratory study of 90 participants, fluency on the Alternative Uses Task was higher when ideas were written down than when they were spoken aloud.Ning Hao et al., Frontiers in Psychology, 2015
132 first gradersIn a 2016 study of 132 Italian first graders, spelling words and pseudo-words to dictation was more difficult than spontaneous spelling of words, and this held for low achievers and children with spelling impairments too.Lucia Bigozzi et al., PLoS ONE, 2016
49.2% of office-day time on EHR and desk workIn a 2016 time and motion study, physicians spent 27.0% of their office-day time with patients in direct face time and 49.2% on EHR and desk work.Christine Sinsky et al., Annals of Internal Medicine, 2016
55 subjectsIn a 2019 laboratory study of 55 people, the speech transmission index, which measures how intelligible background speech is, had a significant effect on performance decrement, mental load and discomfort.Krist Kostallari et al., arXiv (eess.AS), preprint, 2019
18.8 percent word error rateIn a 2024 baseline test, the OpenAI Whisper model had a word error rate of 18.8 percent on whispered speech.Aref Farhadipour et al., arXiv (Electrical Engineering and Systems Science, Audio and Speech Processing), 2024
In a corpus of task-oriented conversations, disfluency rates were higher when speakers discussed abstract figures, which the authors read as an increase in planning difficulty.Bortfeld H et al., Language and speech, 2001
In a longitudinal study, users became more efficient at dictation, navigation and error correction as they gained experience.Jinjuan Feng et al., Interacting with Computers, 2005
In a meta-analysis of 94 studies, procedures that asked people to describe or explain their thoughts and actions changed their performance, unlike plain thinking aloud.Mark C. Fox et al., Psychological Bulletin, 2011
65 percent of dictation time spent correctingIn a preliminary 2001 study of dictation with a speech recognizer, users spent over 65 percent of their time correcting errors and less than 35 percent dictating.Andrew Sears, ERCIM News, No. 46, 2001
8.12 percent word error rateIn a preliminary Wall Street Journal recognition task, wideband (16 kHz) speech gave a word error rate of 8.12% with the same recognizer architecture.Kehuang Li et al., Interspeech 2015, International Speech Communication Association (ISCA), 2015
In a randomized trial of adults with rheumatoid arthritis, spoken disclosure of stressful experiences led to faster walking speed at 3 months, and reduced pain, swollen joints and physician-rated disease activity at 6 months.Mark A Lumley et al., Pain, 2011
In a study that re-recorded read speech with a distant microphone, recognition performance was dramatically reduced compared with close-talking speech.arXiv preprint, 2017
4.8 minutes per visit on notesIn adjusted models of Epic data, physicians who mainly used transcription and dictation spent less time on notes per visit than physicians who mainly used SmartTools, at an average of 4.8 minutes per visit.Journal of General Internal Medicine, 2023
50 percentIn an in-lab study, participants missed identifying over 50 percent of speech recognition errors when listening to audio output of the recognized text.Jonggi Hong and Leah Findlater, CHI 2018 Extended Abstracts (ACM), 2018
In an oral-history archive with noisy, elderly and accented speech, signal-to-noise ratio and syllable rate were the two dominant factors explaining the overall word error rate.Olivier Siohan et al., Interspeech 2004 (ISCA), 2004
In Apple's Voice Control, Dictation mode, which types every word that is not a command, is on by default.Apple Support, Mac User Guide (macOS 27 version of the page), 2026
In Apple's Voice Control, saying Command mode makes Voice Control respond only to commands, and words and characters that are not commands are ignored rather than entered as text.Apple Support, Mac User Guide (macOS 27 version of the page), 2026
In Apple's Voice Control, you act on a text item such as a word by saying a command like Delete followed by the item's number.Apple Support, Mac User Guide (macOS 27 version of the page), 2026
1997 yearIn April 1997, Dragon Systems unveiled NaturallySpeaking, the first general-purpose, large-vocabulary continuous speech recognition product.Susan Hura, Speech Technology Magazine, 1998
In Charlotte Mason's dictation lesson, a child of eight or nine first prepares a paragraph, and older children prepare a page or two or three pages.Charlotte M. Mason, Project Gutenberg, eBook 71087, 1886
7.4 percent of wordsIn clinical notes dictated in 2016, the draft produced by speech recognition had an error rate of 7.4 percent, that is 7.4 errors per 100 words.Li Zhou et al., JAMA Network Open, 2018
In Feng, Karat and Sears's study, navigation and error correction took a big share of the dictation task during the early stages of use.Jinjuan Feng et al., Interacting with Computers, 2005
In Google Cloud Speech-to-Text, the profanity filter does not run unless the profanityFilter setting is true.Google Cloud documentation, n.d.
In Google Docs, you start voice typing by opening a document in a supported browser, choosing Tools then Voice typing, and clicking the microphone box that appears.Google (Docs Editors Help), 2026
In Gould, Conti and Hovanyecz's 1983 experiments, some versions of an imperfect listening typewriter could be at least as good as traditional handwriting and dictating for composing letters, even on first use.John D. Gould et al., Communications of the ACM, 1983
In IBM's 1978 study, novice dictators learned within a few hours to dictate letters as fast and as well as they wrote them, although they did not believe they were doing that well.John D. Gould and Stephen J. Boies, Science, 1978
In Kellogg's 1988 experiments, composing a rough first draft instead of a polished draft did not improve the quality of the writing.Ronald T. Kellogg, Journal of Experimental Psychology: Learning, Memory, and Cognition, 1988
In Keyboard settings on a Mac, the option for what the fn or Globe key does can be set to Start Dictation, which starts dictation when you press that key twice.Apple (Mac User Guide, macOS 27), 2026
7 words in a turnIn large corpora of conversational telephone speech in English and Chinese, speaking rate rises rapidly as a speaker's turn grows from one word to seven words, then levels off for medium-length turns.Jiahong Yuan et al., Interspeech 2006 (ISCA), 2006
1990 yearIn March 1990, Dragon Systems introduced DragonDictate, which the company said was the first commercially viable typewriter that works by human speech.Deseret News, Deseret News, 1990
In Mason's dictation lesson, the child prepares alone by looking at each word they are not sure of, and then seeing it with their eyes shut.Charlotte M. Mason, Project Gutenberg, eBook 71087, 1886
In Mason's dictation lesson, the teacher is advised to be ready with slips of stamp-paper to cover a wrong word, so that its image may be erased as far as possible.Charlotte M. Mason, Project Gutenberg, eBook 71087, 1886
In Mason's dictation lesson, the teacher writes each doubtful word on the blackboard, lets the child look at it until they have a picture, and then rubs the word out.Charlotte M. Mason, Project Gutenberg, eBook 71087, 1886
In Microsoft 365 Dictation, saying 'delete' by itself removes the last word or punctuation mark before the cursor.Microsoft, 2026
In Microsoft 365 Dictation, the Auto Punctuation setting can be turned on or off, but only where it is available for the language you chose.Microsoft, 2026
In Microsoft 365 Dictation, the formatting commands include 'capitalize' followed by a word or phrase, 'capitalize from' a point to another point, and 'capitalize last' a number of characters, words, sentences or paragraphs.Microsoft, 2026
In Nuance's Dragon for PC help, you open the Spelling Window by saying "Spell That" and then spelling the word or phrase, which lets you add a word by its letters.Nuance Communications, Dragon for PC help, undated
In online studies, adding a pause between words improved how well people identified speech recognition errors by ear, compared with no pause.Jonggi Hong and Leah Findlater, CHI 2018 Extended Abstracts (ACM), 2018
In schools, a dictation is a test in which a teacher reads a passage aloud and students learning a language write it down.Cambridge University Press, Cambridge University Press, 2026
In supported languages, Dictation on iPhone inserts commas, periods and question marks as you speak, without a spoken command.Apple Inc., 2026
In the 1992 study, participants made significantly more memorization errors when using speech than the mouse to activate commands.Lewis Karl et al., Unpublished manuscript, University of Maryland (PDF hosted by Ben Shneiderman's site), 1992
In the 2010 habit study, performing the behavior more consistently was associated with a better model fit.Phillippa Lally et al., European Journal of Social Psychology, 2010
18 to 254 daysIn the 2010 habit study, the time to reach 95% of the automaticity ceiling ranged from 18 to 254 days between people.Phillippa Lally et al., European Journal of Social Psychology, 2010
54.1 percent lower average failure rateIn the 2021 SmartEdit study, the authors' editing techniques lowered the average failure rate by 54.1 percent compared with descriptive commands or re-speaking.Jiayue Fan et al., Proceedings of UIST 2021, ACM Symposium on User Interface Software and Technology, 2021
1.30 minutes per visit less than SmartToolsIn the adjusted comparison, physicians who mainly used transcription and dictation spent 1.30 minutes less per visit on notes than physicians with predominant SmartTools use.Journal of General Internal Medicine, 2023
In the AirPods settings on a Mac, Apple's Automatically Switch AirPods option lets either of your AirPods act as the microphone.Apple, Apple Support (AirPods User Guide), n.d.
In the browser speech recognition interface, the continuous setting defaults to false, which means each recognition returns a single result rather than continuous results.MDN contributors, MDN Web Docs (Mozilla), 2026
3.2 times the close-talk word error rateIn the Ravanelli et al. 2017 figures, a single distant microphone gave a word error rate about 3.2 times the close-talking rate. The two conditions are not identical, so this is a rough ratio.Calculated from other facts
In the same 1988 experiments, preparing a written outline before writing led to higher-quality documents than not outlining, as rated by judges.Ronald T. Kellogg, Journal of Experimental Psychology: Learning, Memory, and Cognition, 1988
32 percent of time navigating within the documentIn the same 2001 study, users spent over 32 percent of their time moving from one place to another within the document.Andrew Sears, ERCIM News, No. 46, 2001
In the same 2002 study, essay length did not differ significantly between speech recognition, human transcription and keyboarding.Richard Conrad Snider, Virginia Tech (VTechWorks repository), 2002
6 characters per reportIn the same 2013 study, experienced radiologists corrected 6 characters per report.Tomi A. Kauppinen et al., Journal of Digital Imaging, 2013
52 words per minuteIn the same controlled study on an iPhone, English text entered on the built-in QWERTY keyboard reached 52 words per minute, against 153 words per minute by speech.Sherry Ruan et al., Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), 2018
8.67 percent word error rateIn the same preliminary Wall Street Journal task, narrowband (8 kHz) speech gave a word error rate of 8.67%.Kehuang Li et al., Interspeech 2015, International Speech Communication Association (ISCA), 2015
In the same randomized trial of adults with rheumatoid arthritis, written disclosure of stressful experiences had minimal effects compared with the combined control groups.Mark A Lumley et al., Pain, 2011
3.7 percentIn the same study, close-talking read speech recorded in a studio had a word error rate of 3.7 percent with DNN models.arXiv preprint, 2017
43 words per minuteIn the same study, Mandarin Chinese text entered with the built-in Pinyin keyboard reached 43 words per minute, against 123 words per minute by speech.Sherry Ruan et al., Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), 2018
In Voice Control's Dictation mode, any words you say that are not Voice Control commands are entered as text.Apple Support, Mac User Guide (macOS 27 version of the page), 2026
In voice writing, the mask stops the reporter's voice from being heard in the room while the reporter repeats the testimony into the recorder.Wikipedia, 2026
In Windows 11, apps can use the microphone only when Microphone access is turned on under Privacy & security, App permissions, Microphone.Microsoft, Microsoft Support, n.d.
In Windows Voice Access, saying "Correct that" corrects the selected text or the last dictated text.Microsoft Support, 2026
In Windows Voice Access, saying "Delete that" deletes the selected text or the last dictated text.Microsoft Support, 2026
In Windows Voice Access, saying "Select that" selects the last text you dictated.Microsoft Support, 2026
In Windows Voice Access, saying Type or Dictate before a voice access command inserts that command as text in a text box.Microsoft Support, 2026
In Windows voice typing, if a word or phrase is selected, any of the delete commands removes the selected text.Microsoft, 2026
In Windows voice typing, you dictate an uppercase letter by saying 'uppercase' before the letter.Microsoft, 2026
Alt+` keyboard shortcutIn Word for Windows, you start dictation from the keyboard by pressing Alt+H, D or Alt + the backquote key (`).Microsoft Support, 2026
1837 yearIsaac Pitman published Stenographic Sound-Hand in 1837, introducing Pitman shorthand.History of Information, n.d.
Kellogg's 2001 experiments found that planning, translating ideas into sentences and reviewing compete for one general-purpose working memory resource.Ronald T. Kellogg, American Journal of Psychology, 2001
Kraal and colleagues say novice dictation users keep correcting errors by voice, which shows an incorrect mental model of dictation.Ben Kraal et al., Proceedings of the 9th Australian International Conference on Speech Science and Technology, Melbourne, 2002, 2002
Legal Futures says digital dictation recordings can be encrypted in real time, then uploaded to a cloud platform or routed by local network, FTP server or email for transcription.Legal Futures, 2024
30 secondsMac Dictation stops automatically when it detects no speech for 30 seconds.Apple Inc., Apple Support, Mac User Guide, n.d.
Mac Dictation underlines ambiguous text in blue, so you can click the underlined word and choose an alternative.Apple Support, n.d.
1885 yearMark Twain had experimented with dictation as a method of composition as early as 1885.Mark Twain Project, Bancroft Library, University of California, Berkeley, Mark Twain Project, 2010
Mason described the common practice of dictation as a teacher who dictates clause by clause and underlines the spelling errors, sometimes in red ink, for the children to correct.Charlotte M. Mason, Project Gutenberg, eBook 71087, 1886
Medical transcriptionists typically need postsecondary education that leads to a certificate.U.S. Bureau of Labor Statistics, 2025
Merriam-Webster defines dictation as the act or manner of uttering words to be transcribed, and also as the material that is dictated or transcribed.Merriam-Webster editors, Merriam-Webster, 2026
Microsoft describes Azure Speech as a service that provides speech to text, text to speech and other capabilities together.Microsoft, Microsoft Learn, 2026
Microsoft lists transcribing calls in real time or processing a batch of recorded calls as a use case of its Azure Speech service.Microsoft, Microsoft Learn, 2026
5.9 percent error rate (professional transcribers, Switchboard)Microsoft Research reported in 2016 that its system reached human parity, and professional transcribers had a 5.9 percent error rate on the Switchboard portion.W. Xiong et al., arXiv (Microsoft Research), 2017
Microsoft says corrections cannot be made fully with voice in Word dictation, so some fixes still need suggestions or the keyboard.Microsoft Support, 2026
Microsoft says desktop apps have their own setting, Let desktop apps access your microphone, which must be turned on.Microsoft, Microsoft Support, n.d.
Microsoft says Dictation in Microsoft 365 uses speech-to-text with a microphone and a reliable internet connection.Microsoft, 2026
Microsoft says Fluid dictation in voice typing on Copilot+ PCs automatically corrects grammar, punctuation and filler words as you speak.Microsoft Support, 2026
Microsoft says Fluid dictation is automatically disabled in secure fields such as passwords or PINs.Microsoft, Microsoft Support, Accessibility, n.d.
Microsoft says Fluid dictation, a Voice Access feature on Copilot+ PCs, automatically corrects grammar, punctuation and filler words as you speak, which reduces the need for manual editing.Microsoft Support, 2026
Microsoft says that to use voice typing on Windows, you need to be connected to the internet, have a working microphone, and have your cursor in a text box.Microsoft, 2026
Microsoft says voice typing converts speech to text using online speech recognition.Microsoft, Microsoft Support, n.d.
Microsoft says voice typing is not available when the current language is not supported, and tells you to install a supported language.Microsoft, Microsoft Support, n.d.
Microsoft says voice typing needs microphone access, which you turn on under Privacy & security, Microphone.Microsoft, Microsoft Support, n.d.
Microsoft says voice typing on a PC uses online speech recognition, and that you need an internet connection, a working microphone and a cursor in a text box to use it.Microsoft Support, 2026
Microsoft says voice typing on Windows needs a working microphone, an internet connection and the cursor in a text box.Microsoft, 2026
Microsoft says Windows voice typing suppresses audio coming from the PC's speakers, such as Narrator or app playback, so that it listens to your voice.Microsoft Support, 2026
Microsoft says Windows voice typing uses online speech recognition powered by Azure Speech services.Microsoft, 2026
Microsoft says you stop Windows voice typing by saying a command such as Stop listening, or by pressing the microphone button on the voice typing menu.Microsoft, 2026
Microsoft says you switch voice typing languages by changing the input language you use.Microsoft, Microsoft Support, Accessibility, n.d.
Microsoft's accessibility guideline for Xbox describes the text-to-speech chat pattern: a player types what they want to say into a text-entry box, and the text is read aloud to the other players in a synthesized voice.Microsoft, Microsoft Learn (Xbox accessibility guidelines), 2026
Microsoft's documented Xbox speech-to-text feature transcribes what people say in a party chat into an on-screen overlay; it was added in the June 2021 update.Xbox Wire (Microsoft), 2021
Microsoft's English command list for Dictate in Word includes 'dollar sign' for the $ symbol and 'number/pound sign' for the # symbol.Microsoft, 2026
Microsoft's English command table for Dictate in Word lists period, comma, question mark, exclamation mark and new line as spoken punctuation or formatting.Microsoft, 2026
Microsoft's Xbox developer reference defines the speech-to-text source for a game as in-game speech subtitles or transcribed game voice chat, so the console's speech-to-text feature is built around game speech rather than dictation into a text box.Microsoft, Xbox developer documentation (devdocs.xbox.com), 2026
NCH Software says a few foot pedals do not work with its Express Scribe transcription software, and that those pedals cannot be used with it.NCH Software, n.d.
NCH Software says a foot pedal in Express Scribe needs the correct port selected in Express Scribe Settings, one of COM, GamePort or VEC USB.NCH Software, n.d.
1880s decade (late 1880s and early 1890s)Neither Graphophones nor Edison Phonographs sold well for office use in the late 1880s or early 1890s.The Office Museum, n.d.
1.8 timesNew radiologists corrected about 1.8 times as many characters per report as experienced radiologists in the 2013 study.Calculated from other facts
10 weeksNew users' rate of words unknown to the recognizer leveled off at 0.36 per report after about 10 weeks.Tomi A. Kauppinen et al., Journal of Digital Imaging, 2013
NIOSH investigators found a potential for excessive noise exposure with the dictation equipment in hospital transcription, mainly when the volume was turned to its maximum.C. A. Kardous and R. L. Tubbs, National Institute for Occupational Safety and Health (NIOSH), Centers for Disease Control and Prevention, 2006
1.1 percentage pointsNoise compensation experiments on the MALACH corpus reduced the word error rate by 1.1 percentage points.Olivier Siohan et al., Interspeech 2004 (ISCA), 2004
Nuance tells Dragon users to speak clearly and naturally at a normal conversational tone and pace, and to dictate with a consistent speed and volume.Nuance Communications (Dragon help), 2015
Nuance's Dragon for PC help says that to add a word or phrase you specify the word and its spoken form, and you can optionally train Dragon on its pronunciation.Nuance Communications, Dragon for PC help, undated
1 inchNuance's Dragon guidance for a headset microphone is to keep the microphone element about an inch from the side of the mouth, but not touching it.Nuance Communications, Nuance Communications, Dragon for Mac online help, n.d.
Nuance's Dragon guidelines recommend dictating using longer phrases rather than short phrases or individual words, because hearing a whole phrase helps the software predict what you say.Nuance Communications (Dragon help), 2015
Nuance's Dragon guidelines say that, unless the user tells Dragon to stop listening, it records everything it hears, including a phone conversation or a coworker's talk.Nuance Communications (Dragon help), 2015
Nuance's Dragon guidelines tell users to pause briefly before and after dictating punctuation.Nuance Communications (Dragon help), 2015
Nuance's Dragon guidelines tell users to reduce background noise, because Dragon hears and transcribes everything.Nuance Communications (Dragon help), 2015
Nuance's help for Dragon for Mac version 6 says Voice Training helps Dragon learn how you speak, and asks the user to read aloud a story.Nuance Communications, n.d.
OM System says DSS dictation recorders can record in overwrite mode and insert mode, which made the format ideal for dictating.OM System Professional Audio Support, 2026
OM System says its DS2 dictation file format provides file encryption at 128-bit or 256-bit AES.OM System Professional Audio Support, 2026
OM System says the DS2 format supports better sound quality, which it says improves speech recognition results.OM System Professional Audio Support, 2026
Search + D keyboard shortcut to start dictationOn a Chromebook, you turn on Dictation in Settings, Accessibility, under Keyboard and text input, then start it by selecting Speak or pressing Search + d (or Launcher + d).Google (Chromebook Help), 2026
On a Mac running macOS 27, Dictation is turned on in System Settings: choose Keyboard in the sidebar, go to Dictation, turn it on and click Enable.Apple Support, n.d.
On a Mac with Apple silicon, you can type during dictation: the cursor stops pulsing while you type and resumes when you stop, so you can keep dictating.Apple Inc., Apple Support, Mac User Guide, n.d.
On a Mac, 'all caps' formats the next word in capitals, and 'all caps on' formats the next phrase in capitals.Apple Inc., 2026
On a Mac, 'no space on' formats the next phrase without spaces, and 'no space off' resumes the default spacing.Apple Inc., 2026
On a Mac, Apple tells users to select the microphone in Sound settings and raise its input volume if the Mac does not respond to their voice.Apple Inc., Apple Support, macOS User Guide, n.d.
On a Mac, Screen Time can restrict Dictation, because Dictation is one of the apps and features an administrator can turn off.Apple Inc., Apple Support, macOS User Guide, n.d.
On a Mac, the 'caps on' command formats the following phrase in Title Case, and 'caps off' returns to the default lettering.Apple Inc., 2026
On a Mac, the 'numeral' command formats the next phrase as a number, and 'roman numeral' formats it as a Roman numeral.Apple Inc., 2026
On a Mac, turning on Voice Control means it is used for dictation, and standard macOS Dictation is not available.Apple Support, Mac User Guide (macOS 27 version of the page), 2026
On a Mac, you turn Dictation on or off from the Dictation section of Keyboard settings in System Settings.Apple Inc., Apple Support, Mac User Guide, n.d.
On a Meta Quest headset, you dictate by selecting the microphone on the Quest keyboard instead of typing.Meta (Meta Quest Help Center), 2026
On a SpeechMike with a slide switch, you start recording by moving the switch to the REC position and stop by moving it to STOP.Philips Speech (SpeechExec user documentation), 2026
On an iPhone, you turn on Dictation in the Settings app under General, then Keyboard, by turning on Enable Dictation and confirming the prompt.Apple Inc., Apple Support, iPhone User Guide, n.d.
On Android with Gboard, you voice type by tapping a text field, tapping the microphone at the top of the keyboard and speaking when you see "Speak now".Google (Gboard Help), 2026
79.8 macro F1 scoreOn English TED Talks transcripts, the best BERT models reached a macro-averaged F1 of 79.8 for punctuation restoration.Attila Nagy et al., arXiv, 2021
On iPhone, 'All caps' makes the next word uppercase, and 'all caps on … all caps off' makes the enclosed words uppercase.Apple Inc., 2026
On iPhone, 'Cap' capitalizes the next word, and 'caps on' and 'caps off' capitalize the first character of each word in the enclosed words.Apple Inc., 2026
30 secondsOn iPhone, Dictation stops automatically after 30 seconds without speech.Apple Inc., Apple Support, iPhone User Guide, n.d.
On iPhone, saying 'new line' or 'new paragraph' adds a line break during Dictation.Apple Inc., 2026
On iPhone, the 'No space' command, which removes the space between two words, is not available for all languages.Apple Inc., 2026
On iPhone, you add punctuation by saying the name of the mark, for example 'exclamation point'.Apple Inc., 2026
On iPhone, you can spell a word by saying its letters, for example 'S-T-E-P-H-E-N' for a name.Apple Inc., 2026
63.4 percent F1On speech recognition output from the IWSLT 2011 English benchmark, the best reported systems reached 63.4% F1 for question marks.Lucas Rafael Stefanel Gris et al., arXiv (book chapter), 2023
66.3 percent F1On speech recognition output from the IWSLT 2011 English benchmark, the best reported systems reached 66.3% F1 for commas.Lucas Rafael Stefanel Gris et al., arXiv (book chapter), 2023
82.3 percent F1On speech recognition output from the IWSLT 2011 English benchmark, the best reported systems reached 82.3% F1 for periods.Lucas Rafael Stefanel Gris et al., arXiv (book chapter), 2023
25.5 percent word error rateOn the CHiME6 test set, Whisper Large V2 had a word error rate of 25.5 percent.Alec Radford et al., arXiv (eess.AS), 2022
2.7 percent word error rateOn the LibriSpeech Clean test set, Whisper Large V2 had a word error rate of 2.7 percent.Alec Radford et al., arXiv (eess.AS), 2022
76.7 percent F1On the reference (correct) transcript of the IWSLT 2011 English benchmark, the best reported comma F1 was 76.7%.Lucas Rafael Stefanel Gris et al., arXiv (book chapter), 2023
88.9 percent F1On the reference (correct) transcript of the IWSLT 2011 English benchmark, the best reported period F1 was 88.9%.Lucas Rafael Stefanel Gris et al., arXiv (book chapter), 2023
87.8 percent F1On the reference (correct) transcript of the IWSLT 2011 English benchmark, the best reported question mark F1 was 87.8%.Lucas Rafael Stefanel Gris et al., arXiv (book chapter), 2023
13.8 percent word error rateOn the Switchboard test set, Whisper Large V2 had a word error rate of 13.8 percent.Alec Radford et al., arXiv (eess.AS), 2022
680,000 hours of training audioOpenAI's Whisper speech recognition models were trained on 680,000 hours of multilingual and multitask audio with transcripts.Alec Radford et al., arXiv (eess.AS), 2022
OpenWhispr's help page says its dictation can be started by pressing once and stopped by pressing again, or by holding the key while you speak, and that which one you can use depends on your platform.OpenWhispr documentation, 2026
0 to 25 percent faster than novicesPeople with years of dictation experience composed letters between 0 and 25 percent faster than novices, depending on how complex the letter was.John D. Gould and Stephen J. Boies, Science, 1978
1963 yearPhilips first presented its pocket recorder and the compact cassette to the public at the 1963 IFA radio show in Berlin.Philips Museum, n.d.
1967 yearPhilips introduced a mini-cassette for dictation machines in 1967, a year after it built a cassette recorder into a portable radio.Philips Museum, n.d.
Philips says Event mode is the default mode of a SpeechMike.Philips Speech (SpeechMike user manual), 2026
100 milliamperesPhilips says some computers limit the output power of their USB ports to 100 mA, which may not be enough for a SpeechMike to work properly.Philips Speech (SpeechMike user manual), 2026
Philips says that if its troubleshooting table does not solve a SpeechMike problem, you should contact the supporting Philips partner you bought the device from.Philips Speech (SpeechMike user manual), 2026
Philips says that when a computer cannot detect the foot control, you should disconnect the foot control and docking station, reconnect the foot control to the docking station first, and then connect the docking station to the computer.Philips Speech (SpeechMike user manual), 2026
Philips says that when a SpeechMike does not record sound because the input level is too low, you should select it as the default recording device in the Windows sound settings.Philips Speech (SpeechMike user manual), 2026
Philips says that when a SpeechMike is completely without function because of USB power, you should use the supplied power supply unit or connect the SpeechMike through a self-powered hub.Philips Speech (SpeechMike user manual), 2026
Philips says the SpeechMike is the official successor to Nuance's PowerMic 3 and PowerMic 4 dictation microphones.Philips Speech, 2026
Philips says you can change the button assignment for each SpeechMike button mode in its SpeechControl software.Philips Speech (SpeechMike user manual), 2026
6 button modesPhilips' manual for the SpeechMike Premium Air lists six button modes: Event mode, Keyboard mode, Browser mode, Windows speech recognition mode, Dragon for Mac mode and Dragon for Windows mode.Philips Speech (SpeechMike user manual), 2026
Philips' manual says that with a SpeechMike you can control various applications on your computer or in your browser.Philips Speech (SpeechMike user manual), 2026
13.1 minutes per visit on notesPhysicians who mostly used copy/paste spent more time on notes, at an average of 13.1 minutes per visit.Journal of General Internal Medicine, 2023
Plastic dictation belts were later replaced by magnetic tape and then by hard drive dictation machines.Speech Technology magazine, n.d.
2 scribes kept busy at once (at least)Plutarch reports that Caesar practiced dictating letters on horseback while keeping two scribes busy at once, or more.Plutarch, Perseus Digital Library, Tufts University, c. 100
8 distances from the sourcePrasad and Sim built a far-field speech corpus by re-recording Wall Street Journal audio with microphones at eight different distances from the source.Animesh Prasad and Khe Chai Sim, Interspeech 2016, International Speech Communication Association (ISCA), 2016
Radford and colleagues report that, compared with humans, their Whisper speech recognition models approach human accuracy and robustness.Alec Radford et al., arXiv (eess.AS), 2022
Readability studies cited by Lease and Johnson in 2006 found that fillers and speech repairs can be deleted from transcripts without compromising meaning.Matthew Lease and Mark Johnson, Proceedings of NAACL HLT 2006, Short Papers, 2006
25 times fewer errors after reviewReview and editing cut the error rate of dictated clinical notes from 7.4 to 0.3 errors per 100 words, about 25 times fewer.Calculated from other facts
Simply Charlotte Mason describes copywork as giving the student a beautiful model of handwriting to copy, with the focus on handwriting at grades 1 and 2.Simply Charlotte Mason, Simply Charlotte Mason (blog), n.d.
Simply Charlotte Mason describes transcription as the stage after handwriting mechanics, when the child starts to concentrate on the spelling of the passage being copied.Simply Charlotte Mason, Simply Charlotte Mason (blog), n.d.
Simply Charlotte Mason says that after a student is comfortable writing words from memory, the student moves to dictation: preparing a whole passage and writing it all at once as the teacher dictates it.Simply Charlotte Mason, Simply Charlotte Mason (blog), n.d.
Some new radiologists had a very unfavorable learning curve, with error rates that did not fall as expected.Tomi A. Kauppinen et al., Journal of Digital Imaging, 2013
Sony describes PS5 Voice Command (Preview) as a way to use your voice to find or open games, apps and settings, capture gameplay or control media playback, so its purpose is console control rather than typing text into a field.PlayStation Support (Sony Interactive Entertainment), 2026
Sony says that with some devices that have noise cancellation, Voice Command may have trouble understanding you, and it suggests using the microphone on the wireless controller instead.PlayStation Support (Sony Interactive Entertainment), 2026
Sony's Voice Command (Preview) for PlayStation 5 is available only to accounts registered in the United States and the United Kingdom, and it listens and responds only in English.PlayStation Support (Sony Interactive Entertainment), 2026
6.1 minutes per visit on notes (approximate)Taking the adjusted difference from the dictation group, SmartTools users averaged about 6.1 minutes per visit on notes.Calculated from other facts
The 2010 habit study found that missing one opportunity to perform the behavior did not materially affect habit formation.Phillippa Lally et al., European Journal of Social Psychology, 2010
The 2015 study's mediation analysis found that cognitive demand fully accounted for the difference in idea fluency between writing and speaking.Ning Hao et al., Frontiers in Psychology, 2015
The 2016 study found a moderate correlation between spelling to dictation and spontaneous spelling, which the authors read as showing that the two tasks are supported by partially different spelling processes.Lucia Bigozzi et al., PLoS ONE, 2016
57 physicians observedThe 2016 time and motion study observed 57 physicians for 430 hours.Christine Sinsky et al., Annals of Internal Medicine, 2016
The 2017 paper says distant speech recognition still lacks robustness when non-stationary noise and acoustic reverberation are present.arXiv preprint, 2017
The 2019 paper argues that automatic punctuation should favor precision over recall, because readers can add a missing mark in their heads more easily than remove a wrong one.Gyorgy Szaszak, Acta Polytechnica Hungarica, 2019
The 2019 paper proposes and tests a punctuation model built on prosody, the pitch and timing cues of speech, rather than on the words alone.Gyorgy Szaszak, Acta Polytechnica Hungarica, 2019
The 2019 paper says that for dictation systems, where punctuation matters most, a telegraphic style with explicit dictation of the punctuation marks was foreseen.Gyorgy Szaszak, Acta Polytechnica Hungarica, 2019
300 words per minuteThe ability to identify speech recognition errors by ear degrades at higher speech rates, and the study tested 300 words per minute.Jonggi Hong and Leah Findlater, CHI 2018 Extended Abstracts (ACM), 2018
The AS-2700 foot switch is HID keyboard compatible and can be customized with a free Windows configuration tool.OM Digital Solutions (OM System dictation), 2026
The AS-2700 transcription kit is no longer sold as a bundled product, although its software and accessories are still sold separately.OM Digital Solutions (OM System dictation), 2026
The AS-2700 transcription kit's foot switch has three pedals and gives hands-free control.OM Digital Solutions (OM System dictation), 2026
14 percent relative gainThe Aurora standardization tests, as cited by Bauer et al., found a 14% average relative gain in recognition of noisy speech sampled at 16 kHz instead of 8 kHz.Patrick Bauer et al., EUSIPCO 2014, European Signal Processing Conference, 2014
24 fold improvement in median turnaroundThe authors describe a 24-fold improvement in the median radiology report turnaround after speech recognition was introduced.Luciano M. Prevedello et al., Journal of the American College of Radiology, 2014
The authors state that for distant speech the performance of automatic speech recognition is dramatically reduced compared with the close-talking case.Mirco Ravanelli et al., arXiv (preprint, eess.AS), 2017
The authors trace the speaker-group gap to the acoustic models used by the ASR systems.Allison Koenecke et al., Proceedings of the National Academy of Sciences, 2020
The authors write that it seems accepted that speech noise in open-plan offices is the main source of discomfort for employees.Krist Kostallari et al., arXiv (eess.AS), preprint, 2019
40 words per minute (fast typists)The average typing rates of the study's participants on plain text were 23, 35 and 40 words per minute for the slow, average and fast typists.Bernhard Suhm et al., Proceedings of CHI 99, ACM Conference on Human Factors in Computing Systems, 1999
10.4 percentage pointsThe best reported comma F1 is 10.4 percentage points lower on speech recognition output than on reference text.Calculated from other facts
8 kHz sample rate for the narrowband codecThe Bluetooth Headset Profile carries voice either at 8 kHz with the CVSD codec or at 16 kHz with the mSBC wideband codec.Audio Precision, Audio Precision, technical news, 2012
The Cambridge Business English Dictionary defines dictation as speaking aloud for a person or a machine to record, so the words can be written down.Cambridge University Press, Cambridge University Press, 2026
The Cambridge English Dictionary defines dictate, as a verb, as speaking something aloud for a person or machine to record what is said, so that it can be written down.Cambridge University Press, Cambridge University Press, 2026
The Cambridge English Dictionary defines dictation, as an uncountable noun, as the activity of dictating something for someone else to write down.Cambridge University Press, Cambridge University Press, 2026
The Cambridge English Dictionary defines diction, an uncountable noun, as the manner in which words are pronounced.Cambridge University Press, Cambridge University Press, 2026
The Cambridge English Dictionary gives a second verb sense of dictate: to give orders, or to tell someone exactly what they must do, with total authority.Cambridge University Press, Cambridge University Press, 2026
The Cambridge English Dictionary gives two senses of transcription: a written record of words or music, and the process of transcribing something.Cambridge University Press, Cambridge University Press, 2026
The Cambridge English Dictionary labels dictate as a formal noun, usually used in the plural, meaning an order that should be obeyed, often one that you give to yourself.Cambridge University Press, Cambridge University Press, 2026
The DSS format was designed for digital voice recorders and gives a high compression ratio with little loss of quality.Philips Speech, 2026
The editors of Mark Twain's autobiography write that his autobiographical dictations would form the bulk of the book he called the Autobiography of Mark Twain.Mark Twain Project, Bancroft Library, University of California, Berkeley, Mark Twain Project, 2010
The English voice typing command table in Windows lists 'new line', 'new/next line' and 'new paragraph' as line break commands.Microsoft, 2026
215,207 ambulatory physiciansThe Epic analysis covered 215,207 U.S. ambulatory physicians using the Epic EHR between September 2020 and May 2021.Journal of General Internal Medicine, 2023
The gear icon in the Windows voice typing flyout opens settings that include the voice typing launcher, which offers voice typing when you are in a text box.Microsoft, 2026
39 participantsThe habit curve fitted 62 individuals in the 2010 study, and 39 of them showed a good fit.Phillippa Lally et al., European Journal of Social Psychology, 2010
The HIPAA Security Rule lists encryption and decryption of electronic protected health information as an addressable implementation specification.U.S. Department of Health and Human Services (text reproduced by Cornell Legal Information Institute), 2013
The iPhone command list for Dictation includes period, comma, exclamation point, question mark, dollar sign and parentheses.Apple Inc., 2026
The largest productivity gains in the longitudinal dictation study came from better dictation quality and from using navigation commands.Jinjuan Feng et al., Interacting with Computers, 2005
The Mac command table maps the spoken phrase 'question mark' to a question mark, and 'quote' and 'end quote' to opening and closing quotation marks.Apple Inc., 2026
The Mac formatting commands include 'new line', which starts a new line, and 'new paragraph', which starts a new paragraph.Apple Inc., 2026
The MALACH study observed no evidence that accent or speaker age affected recognition performance.Olivier Siohan et al., Interspeech 2004 (ISCA), 2004
$40,410 median annual wage, May 2025The median annual wage for medical transcriptionists was $40,410 in May 2025.U.S. Bureau of Labor Statistics, 2025
225 words per minuteThe NCRA's Registered Professional Reporter skills test sets the testimony and question-and-answer leg at 225 words per minute.National Court Reporters Association (NCRA), n.d.
The Oxford Advanced Learner's Dictionary defines dictate, as a verb, as saying words for somebody else to write down or to be recorded.Oxford University Press, Oxford University Press, 2026
The Oxford Advanced Learner's Dictionary defines diction, an uncountable noun, as the way that somebody pronounces words.Oxford University Press, Oxford University Press, 2026
The Oxford Advanced Learner's Dictionary gives a specialist sense of diction: the choice and use of words in literature.Oxford University Press, Oxford University Press, 2026
The revised Ruan paper says upper-bound speech recognition performance has greatly improved compared with prior systems.Sherry Ruan et al., Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), 2018
The Ruan authors say that results in non-laboratory settings still need study, for both speech and keyboard entry.Sherry Ruan et al., Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), 2018
125 to 150 words per minuteThe same authors put people's normal speaking speed at 125 to 150 words per minute.Jinjuan Feng et al., Interacting with Computers, 2005
The same meta-analysis found that describing or explaining one's thoughts and actions changes performance: it was higher than in silent control conditions.Mark C. Fox et al., Psychological Bulletin, 2011
The Whisper paper says its evaluation audio covers recording conditions including telephone and VoIP calls and meetings.Alec Radford et al., arXiv (eess.AS), 2022
1652 year (blindness begins)The Wikipedia article on Paradise Lost says Milton, blind from 1652, wrote the poem entirely through dictation with the help of amanuenses and friends.Wikipedia contributors, Wikipedia, n.d.
The Wikipedia article on Paradise Lost says the image of Milton dictating the poem to his daughters became a popular subject for paintings, especially in the Romantic period.Wikipedia contributors, Wikipedia, n.d.
The Windows voice typing settings include Automatic punctuation, which inserts punctuation based on what you say.Microsoft, 2026
The Windows voice typing symbol table lists the spoken phrase 'question mark' for the ? symbol.Microsoft, 2026
1624 year of first known useThe word dictation comes from Latin dictare, to say aloud words to be transcribed by another, and its first known use in English dates to 1624.Merriam-Webster editors, Merriam-Webster, 2026
Thomas Edison believed the phonograph's primary use would be in business as a dictation machine, not entertainment.ETHW (IEEE History Center), 2019
Throughout most of the twentieth century, business people recorded voice letters, and a secretary typed them up from the recordings.ETHW (IEEE History Center), 2019
225 words per minuteTo earn the RPR, a court reporter must transcribe dictation at 180, 200 and 225 words per minute with 95 percent accuracy on each test.National Court Reporters Association, National Court Reporters Association, 2026
95 percent accuracyTo pass each leg of the NCRA skills test, a candidate must have 95 percent accuracy.National Court Reporters Association (NCRA), n.d.
To stop Windows voice typing, you say a command such as "Stop listening" or press the microphone button on the voice typing menu.Microsoft, 2026
To turn off automatic punctuation on iPhone, you go to Settings, then General, then Keyboard, and turn off Auto-Punctuation.Apple Inc., 2026
2014 model yearToyota's 2019 service bulletin lists, among the faults fixed by its Entune software update, a voice command system that freezes when a voice command is used immediately after the ignition is turned on, on Entune Premium Audio units in 2014 model year Corolla and RAV4 vehicles.Toyota Motor Sales, U.S.A., technical service bulletin, revision 1, 2019
1906 yearTwain began almost daily dictations to a stenographer in January 1906.Mark Twain Project, Bancroft Library, University of California, Berkeley, Mark Twain Project, 2010
6 decibels per doubling of distanceUnder ideal conditions with no reflections or background sound, a sound level drops by 6 dB for every doubling of the distance from the source.Simon Fraser University, Sonic Studio, Simon Fraser University, Sonic Studio acoustics handbook, n.d.
Unity developer documentation says audio over a Bluetooth voice connection is of lower quality, similar to a standard phone call, and mono, while A2DP audio is high quality and stereo.Unity Technologies, Unity Vivox developer guide, Android, n.d.
40 percent lexical density (threshold)Ure (1971), as cited by Ishikawa (2015), put the lexical density of speech below 40 percent and that of writing above it.Shin'ichiro Ishikawa, Procedia - Social and Behavioral Sciences, 198 (2015) 202-210, 2015
Vertanen reports that when dictating with a large-vocabulary recognizer, correcting errors takes up most of the task time.Keith Vertanen, Interspeech 2006 (ICSLP), Pittsburgh, 2006
2 dictation modesVoice dictation on Meta Quest headsets has two modes: online dictation, only available to users in the United States, and on-device dictation, an opt-in feature available globally.Meta (Meta Quest Help Center), 2026
When a website wants the microphone, Chrome on a computer asks you to choose Allow while visiting the site, Allow this time, or Never allow.Google (Chrome Help), 2026
When automatic punctuation is on, Google Cloud Speech-to-Text also capitalizes the first letter after each period and question mark.Google Cloud, 2026
When Gboard first uses the microphone, Android asks for a recording permission with three choices: While using app, Only this time, or Don't allow.Google (Gboard Help), 2026
When people correct dictation errors by speaking again, the spoken correction is often misrecognized as well, which costs time.Keith Vertanen, Interspeech 2006 (ICSLP), Pittsburgh, 2006
9.4 timesWhisper Large V2 has about 9.4 times the word error rate on CHiME6 that it has on LibriSpeech Clean.Calculated from other facts
Wikipedia defines a noise-canceling microphone as a microphone designed to filter ambient noise.Wikipedia, the free encyclopedia, n.d.
Wikipedia defines shorthand as an abbreviated symbolic writing method that increases the speed and brevity of writing compared with longhand.Wikipedia, the free encyclopedia, n.d.
Wikipedia describes voice writing as a transcription method used for court reporting, medical transcription, CART and closed captioning.Wikipedia, 2026
1888 yearWikipedia says Gregg shorthand was first published in 1888 by John Robert Gregg, and that it largely superseded Pitman shorthand in the United States and some other parts of the world.Wikipedia, the free encyclopedia, n.d.
Wikipedia says that noise-canceling microphones of this design have at least two sound ports, a front port aimed at the desired sound and a more distant port.Wikipedia, the free encyclopedia, n.d.
Windows 11 Voice Access controls the PC and writes text by voice without an internet connection.Microsoft Support, 2026
Windows Voice Access can delete the previous or next word, using commands such as Delete previous word and Delete last word.Microsoft Support, 2026
Windows voice typing has a settings option called Filter profanity, which lets you block or allow typing out profanity.Microsoft, 2026
Windows voice typing has a wait-time setting that controls how long it waits before carrying out a voice command, which Microsoft says gives more accurate recognition whether you speak slowly or quickly.Microsoft Support, 2026
Windows voice typing has an Automatic punctuation setting, which lets it insert punctuation based on what you say.Microsoft Support, 2026
Windows voice typing lets you dictate most numbers and punctuation by saying the number or punctuation character.Microsoft, 2026
Windows voice typing needs an internet connection, a working microphone, and the cursor in a text box.Microsoft, 2026
Windows voice typing uses online speech recognition that Microsoft says is powered by Azure Speech services.Microsoft, 2026
Windows voice typing uses spoken commands such as "delete that" and "select that" to edit the text you just dictated.Microsoft, 2026
12.0 percent word error rateWith a single distant ceiling microphone in an apartment and real recordings, the DNN recognizer's word error rate was 12.0%.Mirco Ravanelli et al., arXiv (preprint, eess.AS), 2017
With Google Cloud Speech-to-Text's profanity filter on, the transcript keeps only the first letter of a profane word and replaces the rest with asterisks, for example f***.Google Cloud documentation, n.d.