Transcription
Transcription: How to Turn Audio and Video Into Text
Transcription turns recorded speech into text, and for most recordings an AI transcriber now produces a usable draft in minutes, for cents per audio hour, on your own device or in the cloud. What you end up with depends on the audio, the transcription style and the review, so pay for human work, full verbatim, timestamps and speaker labels only where the use (legal records, research, captions, publication) demands them.
Updated October 16, 202620 min read53 cited facts, each traced to its source
442times
At $0.0045 per minute an hour of audio costs about $0.27 through OpenAI's API, against about $119.40 at Rev's $1.99 per minute for human transcription, roughly 440 times more.
How it is calculated
$1.99per minute
Rev's price page lists human transcription with a 99 percent plus accuracy guarantee in 12 hours or less at $1.99 per minute.
Rev, Rev, 2026
$0.0045per minute
OpenAI's price list shows its gpt-transcribe model at $0.0045 per minute and gpt-4o-transcribe at $0.006 per minute.
OpenAI, OpenAI, 2026
1percent
In a 2024 study, roughly 1 percent of Whisper audio transcriptions contained entire hallucinated phrases or sentences that did not exist in the audio.
Allison Koenecke et al., arXiv (ACM FAccT 2024), 2024
Guides in this topic
Each guide takes one part of the subject in depth.
- How to Transcribe Audio to TextAny recording can become text in minutes: run it through a speech recognition tool on your own device or in the cloud, then proofread the draft against the audio. Pay a human transcriptionist only when the audio is hard to hear or the stakes are high.
- Meeting Transcription: How It Works and What to UseA meeting can be transcribed four ways: by the meeting platform, by a note-taker bot that joins the call, by an app that captures the audio on your own computer, or by a recorder in the room. The right one depends on where the meeting happens, who needs the transcript, and whether every participant has agreed to it.
- Zoom, Teams and Google Meet TranscriptionZoom, Microsoft Teams and Google Meet each transcribe meetings, but only when someone turns it on, and Zoom and Google Meet usually need a paid plan. Each saves the transcript in a different place. On a free Zoom account, record the meeting to your computer and transcribe the file yourself.
- Video to Text: Transcribe Videos and YouTubeIf a video already has captions, you can often copy its text instead of transcribing it. If it has none, you transcribe the audio track like any recording, then check the draft against the video. Downloading someone else's video to transcribe it can break the platform's terms and may raise copyright questions.
- Voice Memo to Text: Built-In and Free OptionsMany phone recorders already turn speech into text, including iPhone Voice Memos on iOS 18 or later and Samsung Voice Recorder with Transcript assist on supported Galaxy models. A recording from any other app can be exported as an audio file and transcribed with a file tool. Treat every draft as a first version and check names and numbers against the sound.
- Transcription Software: How to Choose OneThe right transcription software is the one that passes five checks on your own audio: where it processes the recording, how accurate it is on your speakers, whether it labels speakers and times the text, what it exports, and what it costs at your volume. Test two or three tools on ten minutes of your own audio before you pay, because published accuracy figures vary by vendor and by recording.
- Human vs AI Transcription: Accuracy, Cost, SpeedOn clear audio, AI transcription now comes close to human accuracy at a fraction of the cost and turnaround. On noisy, accented, overlapping or certified work, a human pass still wins, so AI first with human review is usually the best value.
- Transcribe Audio With Whisper, ChatGPT or GeminiYes. Whisper is free and open, so you can transcribe a file on your own computer without uploading it, while ChatGPT and Gemini transcribe uploads inside a chat within size and length limits. A chat model may summarize or reword speech, so ask for a verbatim copy and check it against the recording.
- Transcription Jobs: Pay, Skills and Getting StartedGeneral transcription is real work, but it pays by the audio minute, not the hour, so a beginner earns well below the posted rate once listening and checking are counted. AI drafts are moving the job toward editing, and the FTC says legitimate employers do not ask you to pay to start.
- Captions and Subtitles: Formats, Rules and ToolsCaptions are timed text for people who cannot hear the audio, so they name speakers and describe sounds. Subtitles are timed text for people who can hear but do not understand the language. Both are usually delivered as SRT or WebVTT files built from a transcript, and automatic captions are a draft that you should check before you publish.
- How to Transcribe an Interview: Audio to TextTranscribing an interview well starts before the interview, with a clean recording and a decision on transcript style and consent. Let an automatic tool write the first draft, then check every quote against the audio, because error rates depend on the recording and on the speaker.
- Verbatim vs Clean Verbatim TranscriptionFull verbatim records every sound a speaker makes, including filler words, false starts and stutters. Clean verbatim keeps every word that carries meaning and drops the noise, and it is the usual choice for business and media work. Legal records and conversation research usually need full verbatim, so agree the style and its rules in writing before transcription starts.
- Timestamps and Speaker Labels in TranscriptsA timestamp tells you when a word or line starts in the recording, and a speaker label tells you whose voice said it. AI tools add both automatically, but speaker diarization only separates voices into numbered groups: it cannot know their names, and it is least accurate where people talk over each other or sound alike.
- Transcribe and Translate Audio in Other LanguagesAI transcribers are most accurate in widely spoken languages such as Spanish and much less accurate in many others. In a published benchmark, Whisper large-v2 had a word error rate of 3.0 percent in Spanish and 94.8 percent in Yoruba. Translation into English is a second step that needs its own check, and speech that mixes two languages is the hardest case.
Transcription is turning recorded speech into written text. For most recordings the fastest route today is an AI transcriber: upload the file, wait a few minutes, then proofread the draft against the audio. The price gap is large. An hour of audio costs about $0.27 through OpenAI's gpt-transcribe API and about $119 through Rev, a human transcription service, roughly 440 times more 1. That gap is why the usual best value is AI first and a human pass only where the stakes need it.
Three things decide how good the result is, and none of them is the brand of the tool:
- The audio. A close microphone and one speaker at a time beat any model upgrade.
- The style. Verbatim, clean verbatim, timestamps and speaker labels are choices you make before transcribing, and they change what the text is good for.
- The review. AI drafts are accurate on clear speech and still make errors, and a small share of its errors are invented sentences, so names, numbers and quotes need checking against the recording.
This page covers the whole subject: how to transcribe audio, meetings, video and voice memos, what software to use, how human and AI transcription compare, how captions and subtitles differ from transcripts, how timestamps and speaker labels work, other languages, and transcription as a job. Each section gives the short answer with evidence and links to the full guide.
Any recording becomes text in minutes: transcribe with AI, then proofread
The core workflow works with any tool: prepare the file, set the language, transcribe, proofread against the audio, export. The full steps, limits and routes are in our guide to how to transcribe audio to text. The short version:
- Prepare. Use the original recording, not a copy re-saved from a chat app. Trim silence if the file is long.
- Set the language. Automatic detection can fail on short or mixed clips, so choose the spoken language yourself.
- Transcribe. Run the file through an on-device app, a cloud service or a human service (compared below).
- Proofread. Play the audio at normal speed while reading the draft. This takes a fraction of the time that typing it would.
- Export. Plain text, a Word file, or a timed format such as SRT or VTT if you need captions.
Long files have a practical wrinkle. Cloud APIs cap upload size: OpenAI's transcription API accepts files up to 25 MB, and larger recordings must be compressed or split into chunks 2. Split mid-sentence and the model loses context at the seam, so check the joins. Free tiers of web tools add their own minute caps.
The most common questions here have short answers of their own: how to transcribe audio for free, how to transcribe audio offline, how to transcribe low-quality audio, which audio formats can be transcribed and how to transcribe a lecture. If you just want a free file transcriber in your browser, see audio to text.
Three routes: on-device AI, cloud AI and humans
| Route | Where audio goes | Typical cost | Turnaround | Best for |
|---|---|---|---|---|
| Built-in tools (phone recorder, Windows live captions) | Often stays on the device | Free | Live or minutes | Quick personal notes |
| On-device apps (Whisper-based) | Stays on your computer | Free to one-time or subscription | Minutes to a bit longer than the audio, depending on hardware | Private recordings, offline work |
| Cloud AI services and APIs | Uploaded to the provider | Cents per audio hour via API; subscriptions for apps | Minutes | Volume, speaker labels, integrations |
| Human services | Sent to a transcription company | About $1.99 per minute at one large provider 3 | Hours to days | Legal, publication, hard audio, certified work |
On-device AI is now practical on a laptop. OpenAI released Whisper, a general-purpose speech recognition model trained on a large dataset of diverse audio, under the MIT License 4. Apps such as Buzz describe themselves as transcribing and translating audio offline on a personal computer 5. Yap, which we make, is an on-device dictation app for Mac and Windows, and it is one option among many; see offline dictation on Mac and the download page if that is what you need.
Audio quality decides accuracy more than the tool you pick
The same recognizer can be near-perfect on one recording and poor on another. Whisper's own paper shows the spread on a single model, Large V2: a word error rate of 2.7 percent on the clean LibriSpeech read-speech set 6, 13.8 percent on the Switchboard telephone-conversation set 7 and 25.5 percent on the CHiME6 test set 8. (Word error rate, or WER, counts the errors against a correct transcript and divides them by the total number of words in that transcript 9.)
Distance from the microphone matters just as much. In one 2017 speech recognition study, decoding close-talking studio sentences gave a word error rate of 3.7 percent 10, while a single distant ceiling microphone in an apartment gave 12.0 percent 11. The setups were not identical, so treat the ratio of about three as rough 12, but the direction is consistent with everyday experience: a recorder in the middle of a table loses to a phone held near the speaker.
Practical rules that follow: put the microphone within a hand-width or two of each speaker, record in a quiet room, avoid Bluetooth headsets for the main recording (a Bluetooth voice connection is lower quality and mono, unlike high-quality stereo audio 13), and ask people not to talk over each other. How to rescue a poor recording is covered in how to transcribe low-quality audio. For more on how recognition engines are built and measured, see our overview of speech recognition.
Voice memos and phone recordings often transcribe themselves
Your phone's recorder may already do the job. A university help article says iOS and iPadOS 18 added on-device transcription of recordings 14, and that on an iPhone you open a recording, choose Edit Recording and tap the speech bubble icon to generate a transcript 15. Apple says Voice Memos transcription on iPad is available on supported models in English and nine other languages 16. Pixel Recorder and Samsung Voice Recorder offer similar features on supported models and languages.
Anything else can be exported as an audio file and transcribed like any recording. Chat-app voice messages have more limited transcripts of their own. Whatever the source, treat a voice memo transcript as a draft and check names and numbers against the audio. The device-by-device map, with export steps, is in our guide to voice memos and voice notes to text. Quick answers: can Apple Notes transcribe audio, how to transcribe voice recordings on Android, how to transcribe voice messages and how to transcribe a phone call. Phones are also covered in our overview of phones and tablets.
Live captions are a different feature. Windows 11 live captions process audio and generate captions on the device, with nothing sent to the cloud or to Microsoft 17, but they are gone when the audio ends. Apple says its Live Captions accuracy may vary and should not be relied on in high-risk or emergency situations 18.
Meetings: the platform, a bot, a botless app or a recorder in the room
A meeting can be transcribed four ways: the platform's own transcript, a note-taker bot that joins the call, a botless app that captures your device's audio, or a recorder in the room. The right one depends on where the meeting happens, who must see the transcript and whether everyone has agreed. Our guide to meeting transcription compares them, including the difference between a transcript (the record of what was said) and an AI summary (an interpretation of it).
Zoom, Microsoft Teams and Google Meet all transcribe meetings themselves, mostly on paid business plans and usually only when the organizer turns it on. Each stores the result somewhere different. The exact steps and plan requirements are in the guide to Zoom, Teams and Google Meet transcription. Two details from official documentation matter for everyone:
- Teams live transcription appears in real time and includes each speaker's name and time stamp 19, and all participants see a notification that the meeting is being transcribed 20.
- Google Meet told attendees in 2022 that they would see a notification before they join when transcripts are active 21, and saves the transcript as a Google Doc.
That notice is not a legal substitute for consent. Recording and consent laws differ by place, and some require everyone's agreement, so the meeting guide covers consent and local, open-source options that keep audio on your machine. Free-plan users can record locally and transcribe the file. Short answers: meeting transcription without a bot, free meeting transcription, turning a transcript into minutes, transcribing an in-person meeting, whether AI transcription devices are worth it, how to enable transcription in Zoom, how to download a Zoom transcript, whether Zoom can transcribe without recording and whether Google Meet transcribes for free. For the workplace side, see our overview of dictation at work and of voice privacy.
Video to text: copy existing captions first, transcribe the audio track second
If a video already has captions, copy its transcript instead of transcribing it. YouTube shows a transcript for most videos that have a caption track. If there is none, transcribe the video's audio track like any recording, or use an editor that does it in the timeline: Premiere Pro and DaVinci Resolve both transcribe footage inside the project. Audio playing on your computer can also be routed to a transcriber.
Two cautions. Automatic captions are a draft. And downloading someone else's video to transcribe it raises copyright and terms-of-service questions that the guide to video to text explains. Questions people ask: how to get a YouTube transcript, why a YouTube transcript is not showing, how to transcribe audio in Premiere Pro and whether DaVinci Resolve can transcribe audio. For podcasts and interviews specifically, see how to transcribe a podcast.
Captions and subtitles are timed text, built from a transcript
Captions are timed text for people who cannot hear the audio, so they include sounds and speaker changes. Subtitles are timed text for viewers who do not understand the language. Closed captions can be switched off; open captions are burned into the picture. Both are delivered mostly as SRT or WebVTT files.
A WebVTT file consists of a sequence of text segments, each associated with a time interval called a cue 22. SRT is the simpler older cousin: numbered blocks, each with a start and end time and the text. Both are plain text, so you can fix them in any editor. A transcript becomes captions through alignment, which attaches word timings to the text.
The accessibility rules are specific. Under the Web Content Accessibility Guidelines (WCAG) 2.2, captions for prerecorded audio in video are a Level A requirement 23, and captions for live audio are a Level AA requirement 24. Whether a law applies WCAG to your organization is a legal question the guide only describes, not answers. The research on live captioning is a reminder that automation is not yet a replacement for trained captioners: one 2024 study reports that existing approaches for improving automatic speech recognition still fail to match trained CART captioners 25.
Read the guide to captions and subtitles for the formats line by line and for quality standards. Related questions: captions vs subtitles, what an SRT file is, adding auto captions to YouTube Shorts, whether automatic captions are ADA compliant and how to make subtitles with Whisper. Terms: closed captions, open captions, SRT, WebVTT, SDH and CART. Accessibility more broadly is in our overview of accessibility.
Human versus AI: AI wins on cost and speed, humans on hard audio and certified work
On clear audio AI is close to human accuracy; on hard audio humans still win. Whisper's authors report that, compared with humans, their models approach human accuracy and robustness 26, after training on 680,000 hours of multilingual and multitask audio 27. On clean audio, an independent test shows how far engines can differ: in a 2025 9to5Mac test of one 7-minute-31-second episode, Whisper Large V3 Turbo had a word error rate of 1.5 percent, against 10.3 percent for Apple's speech API 28 29. That is one episode, so it shows the spread between engines, not a general ranking.
An accuracy claim means little without the audio and the counting method
"99 percent accurate" can mean very different things. Ask what audio was tested, whether fillers and punctuation were counted, and whether the reference transcript was itself checked. The same model gave 2.7 percent error on read audiobooks 6 and 25.5 percent on CHiME6 recordings 8 (chart above). Accuracy also varies by speaker: a study of five commercial systems found an average word error rate of 0.35 for Black speakers against 0.19 for white speakers 30. Even in medicine, a 2025 neurosurgery pilot measured a mean word error rate of 1.75 percent for a Whisper-based model and 1.54 percent for Dragon Medical One, a difference that was not statistically significant 31 32 33.
Cost and turnaround
| AI transcription | Human transcription | |
|---|---|---|
| Price per audio minute (examples) | $0.0045 for OpenAI's gpt-transcribe API 34 | $1.99 at Rev 3 |
| Cost of one audio hour | about $0.27 | about $119 1 |
| Turnaround | Minutes | Hours to days (Rev promises 12 hours or less 3) |
| Weak spots | Noise, accents, overlap, specialist terms, invented text | Cost, queue time, fatigue on very long files |
API prices are not what an app subscription charges, and an app adds an editor and storage, so compare your real volume. The software guide, how to choose transcription software, gives the arithmetic for per-minute, subscription and one-time pricing. Our guide to human vs AI transcription goes through the accuracy, cost and turnaround evidence, and the questions are answered in how accurate AI transcription is, how much transcription costs, how long it takes to transcribe an hour of audio, how to get transcription without AI and whether transcription software is generative AI.
Legal, medical and published work still calls for a human pass
Where a transcript is evidence, a medical record or a published quotation, a person must check it. The cheapest route to a near-perfect transcript is AI first, human review second: reviewing a draft takes far less time than typing from scratch, which one oral history guide puts at 4 to 9 hours for each audio hour 35.
Whisper, ChatGPT and Gemini: powerful, but they can paraphrase or invent
Whisper is a free, open model you can run on your own computer through an app or the command line. It comes in sizes: the smallest, tiny, has 39 million parameters and needs about 1 GB of memory 36, small has 244 million and needs about 2 GB 37, and the largest has 1,550 million and needs about 10 GB 38. Pick the size by your hardware and how much accuracy you need. ChatGPT can transcribe audio you upload in a chat on paid plans, with files up to 512 MB 39 40. Gemini can transcribe speech in audio input 41, and NotebookLM can ingest audio as a source.
The risk is not typos. It is invention. A 2024 study found that roughly 1 percent of Whisper transcriptions contained entire hallucinated phrases or sentences that did not exist in the audio 42, and 38 percent of that hallucinated content included explicit harms such as violence or false authority 43. A 2025 study found that when non-speech sounds were played to Whisper, a set of hallucinated phrases appeared frequently 44, so check the non-speech stretches of a recording first. Whispered speech is also harder: a 2024 baseline measured an 18.8 percent word error rate for Whisper on whispered speech 45. The 1 percent study tested Whisper as it stood in 2023 46, and models have been updated since, so use it as a reason to spot-check rather than as a current rate.
Chat assistants add a second problem: asked to transcribe, they may summarize or tidy. Prompt for verbatim output and compare a few passages with the audio. The full guide, transcribe audio with Whisper, ChatGPT or Gemini, has the steps, limits and prompts. Questions: how to use Whisper to transcribe audio, can ChatGPT transcribe audio, can Gemini transcribe audio, can NotebookLM transcribe audio and can Whisper transcribe in real time. The wider topic of AI and voice is in our overview of voice and AI.
Choosing transcription software: five checks matter more than brand rankings
Transcription software comes in four kinds: built-in tools, on-device apps, cloud services and human-backed services. Choose by five checks:
- Where the audio is processed. It decides who else can access it. Windows live captions, for example, keep audio on the device 17.
- Accuracy on your own audio. Test ten minutes of your own recording before you pay.
- Speaker labels, timestamps and an editor linked to the audio. These save more time than a point of raw accuracy.
- Export formats. Plain text, DOCX, SRT, VTT, CSV.
- Pricing fit. Per-minute suits occasional use, subscriptions suit steady volume, one-time purchases suit heavy local use.
The detail, plus manual transcription tools that people still use to type transcripts by hand, is in the guide to transcription software. Brand-specific answers: can Dragon transcribe audio files, can Microsoft Word transcribe audio, can Google Docs transcribe audio and what Express Scribe is. If you want live dictation rather than file transcription, start with our overview of dictation software and dictation, and the guide to choosing a dictation app.
Verbatim or clean verbatim: pick the style before anyone transcribes
Full verbatim keeps every sound: fillers, false starts, stutters, pauses. Clean verbatim keeps every meaningful word and drops the noise. Intelligent verbatim also edits grammar for reading and is the least standardized, because providers define the styles differently. Legal records and conversation research usually need full verbatim; most business, media and meeting transcripts use clean verbatim.
| Style | Keeps | Drops | Typical use |
|---|---|---|---|
| Full (true) verbatim | Everything, including "um", repeats, false starts, laughter | Nothing | Courts, conversation analysis, some linguistics |
| Clean verbatim | Every meaningful word | Fillers, stutters, false starts | Interviews, meetings, media |
| Intelligent (edited) | Meaning | Fillers plus grammar fixes and tightening | Articles, reports, books |
AI transcribers default to clean output, which you may need to undo if you need every filler. Providers also differ in how they offer it: some sell verbatim as an add-on rather than the default. Our guide to verbatim vs clean verbatim shows the same passage in each style and the style-guide conventions for tags and inaudible speech. Questions: what a transcript looks like, how to mark inaudible speech and what transcription QA is. Terms: verbatim transcription and clean verbatim.
Timestamps and speaker labels tie text to the audio and to who said it
Timestamps let a reader jump from any line to the moment in the recording. They come per word, per segment, at fixed intervals or at each speaker change. Speaker diarization works out who spoke when, not who the speakers are: it can tell Speaker 1 from Speaker 2, but cannot know names until you supply them. Teams live transcripts include each speaker's name 19, but most recordings you transcribe afterwards need the names added by hand.
Diarization fails where you would expect: overlapping speech, similar voices and short turns such as "yes" and "right". Name the speakers once, then check every label change against the audio. Word-level timestamps are what make captions and audio-linked editors possible, because they come from alignment of the text to the sound. The guide to timestamps and speaker labels explains the styles and the checks. Questions: how to transcribe audio with timestamps for free, how to transcribe audio with multiple speakers and how often to add timestamps. Terms: speaker diarization, timestamp and forced alignment.
Interviews: record well, choose the style, and keep the audio
Transcribing an interview starts before the interview: a good recording halves the work. Choose the transcription style the analysis needs, let AI draft it, and check every quote against the audio, since a misquote costs more than any time saved. Research interviews add consent and data-protection duties, and the audio, not the transcript, remains the primary record, so keep it. Timestamps and speaker labels make quotes findable again.
The guide to how to transcribe an interview covers journalists and researchers together, including preparing transcripts for coding. Questions: can you use AI transcription for research interviews, how to prepare transcripts for qualitative analysis and how to transcribe a podcast.
Other languages: accuracy falls from well-resourced to low-resource languages
AI transcribers handle dozens of languages, but accuracy falls from widely spoken, well-resourced languages to low-resource ones, and mixed-language speech is the hardest case. Set the spoken language yourself rather than relying on detection, and treat translation as a second step even when one tool does both. Scripts without spaces (Chinese, Japanese) and right-to-left scripts need language-aware checking. For low-resource languages, test on your own audio and plan for a human review.
The Whisper study that gives the English figures above also tested many languages, and the pattern of large gaps between languages is one reason to run your own test. Our guide to transcribing and translating audio in other languages reports measured accuracy by language, and the wider subject is in our overview of languages. Questions: how to translate a voice recording to English, how to transcribe Hindi audio, how to transcribe Japanese, Chinese and Korean audio and whether AI can transcribe African languages.
Transcription as a job: low pay per hour, shrinking typing work, growing editing work
General transcription work is real but low-paid and shrinking as AI drafts take over. Platforms pay per audio minute, so speed and audio difficulty set your hourly earnings, and one estimate puts transcribing an audio hour at 4 to 9 hours 35. The work has shifted from typing to editing AI drafts, and the better-paid work is in legal, medical and specialist transcription, which demands training.
The best official figures are for medical transcriptionists, who listen to recorded dictation and transcribe it into medical reports 47. The US Bureau of Labor Statistics reports a median annual wage of $40,410 in May 2025 48, projects employment to decline 4 percent from 2025 to 2035 49, and still projects about 6,200 openings a year on average 50. It says advances in speech recognition let physicians document patient encounters in real time, which reduces the need for transcriptionists 51. Entry typically requires postsecondary education leading to a certificate 52.
Be wary of offers. The Federal Trade Commission says that if someone asks you to pay to get a job, that is a scam 53. The same rule covers transcription offers: a legitimate employer does not ask you to pay for training, software or a starter kit before you earn. The guide to transcription jobs gives the pay arithmetic, the skills tests check, the standard kit (a foot pedal, closed headphones and playback software) and a scam checklist. Questions: how much transcriptionists make, whether transcription is still in demand, what typing speed you need for transcription, how to become a transcriptionist and whether a transcription job is legit. Terms: transcriptionist, turnaround time and transcription foot pedal. Typing speed itself is covered in our overview of typing speed.
All the guides, in the order you will probably need them
| Guide | Use it when |
|---|---|
| How to transcribe audio to text | You have a file and want text |
| Meeting transcription | You need a record of calls or in-person meetings |
| Zoom, Teams and Google Meet transcription | You need exact steps and plan limits for a platform |
| Video to text | The source is a video or a YouTube link |
| Voice memos and voice notes to text | The source is a phone recording |
| Transcription software | You are choosing a tool |
| Human vs AI transcription | You need to decide on accuracy, cost and speed |
| Whisper, ChatGPT or Gemini | You want to transcribe with a model yourself |
| Transcription jobs | You are considering the work |
| Captions and subtitles | You need SRT or VTT or accessibility compliance |
| How to transcribe an interview | You are a journalist or researcher |
| Verbatim vs clean verbatim | You must agree a style |
| Timestamps and speaker labels | You need time codes or "who said what" |
| Other languages | The audio is not in English |
A short checklist that works for almost every recording
- Decide the use first. Notes for yourself need little; evidence, research or publication need verbatim, timestamps and a human check.
- Record close and clean. It is the cheapest accuracy gain available.
- Choose privacy deliberately. Know whether audio leaves your device, and tell people when you record them.
- Start with AI. It costs cents per audio hour and returns in minutes.
- Proofread names, numbers and quotes against the audio. Invented text and silent errors are rare, but they happen.
- Pay a human only where it pays back: hard audio, certified or legal work, and anything you will publish under someone else's name.
- Keep the recording. The audio is the primary record; the transcript is a derivative.
Transcription is a mature, cheap technology wrapped around a small number of real decisions. Make those decisions on purpose, and the rest is a few minutes of waiting and a careful read.
Evidence
Every number on this page, with where it comes from. Open a fact to see the exact words in its source.
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.
- 11.
- 12.
- 13.
- 14.
- 15.
- 16.
- 17.
- 18.
- 19.
- 20.
- 21.
- 22.
- 23.
- 24.
- 25.
- 26.
- 27.
- 28.
- 29.
- 30.
- 31.
- 32.
- 33.
- 34.
- 35.
- 36.
- 37.
- 38.
- 39.
- 40.
- 41.
- 42.
- 43.
- 44.
- 45.
- 46.
- 47.
- 48.
- 49.
- 50.
- 51.
- 52.
- 53.
Questions people ask
- Are AI transcription devices and voice recorders worth it?Usually not. A dedicated AI recorder is worth buying only if you hold many in-person meetings, want the audio kept off your phone and laptop, and accept that the vendor's price, plan and support can change. For video calls, the platform transcript or a headset on your own computer is usually enough.
- Are automatic captions good enough for accessibility rules?No rule we read names automatic captions as compliant or non-compliant on their own. The ADA web rule for state and local governments points to WCAG 2.1 Level AA, which requires captions, so a machine draft needs checking to give the same access as correct captions. This is general information, not legal advice.
- Are transcription courses and certification worth it?For general transcription, a course adds little, because platforms test you directly with a sample. A credential is worth it when the work requires one, such as the AAERT's Certified Electronic Transcriber for legal transcripts, or when employers in medical work may prefer formal training.
- Can AI transcribe Zulu, Xhosa and other African languages?Yes, with a caveat. Google Cloud's speech service and Meta's Omnilingual ASR both list Zulu and Xhosa, but OpenAI's Whisper does not. The only independent figure we found for these languages is for isiXhosa child speech, so test any tool on your own recordings before you rely on it, and plan a human review for anything that matters.
- Can Apple Notes transcribe audio recordings?Yes. In the Notes app on an iPhone or iPad with iOS or iPadOS 18 or later, you can record audio inside a note and show its transcript by tapping the chat bubble icon. If the icon is missing, the phone may not meet the minimum requirements, or the language may not be supported.
- Can ChatGPT transcribe audio files?Yes, on a paid ChatGPT plan. You can attach an audio file of up to 512 MB and ask for its speech as text, but the Free plan does not include audio uploads. A chat model may summarize or reword what it hears, so ask for a word-for-word copy and check it against the recording.
- Can DaVinci Resolve transcribe audio for free?Probably not. Blackmagic Design labels its improved transcription in DaVinci Resolve 20 as a Studio feature, so check the right-click menu in your edition first. If the free edition lacks it, transcribe the audio with a separate tool and check the draft against the recording.
- Can Dragon transcribe audio files?Yes. Dragon can transcribe an audio file you already have. On Windows, Dragon 15 has a Transcribe Recording menu for this. It works best with a recording of one voice: when the speaker is someone else, you correct the text by hand, and it is a poor fit for interviews with several speakers.
- Can Gemini transcribe audio?Yes. You can send an audio file to a Gemini model and ask for its speech as text, and Google's Gemini API also has Gemini 3.5 Transcribe, a dedicated speech-to-text model for files and a live model for streams. A general chat reply may summarize the audio rather than copy it, so ask for a word-for-word transcript and check it against the recording.
- Can Microsoft Word transcribe audio files?Yes, but only with Transcribe, which is separate from Dictate. Microsoft says the feature is currently available only in Word for Microsoft 365 on Windows in commercial tenants, and it accepts uploaded .wav, .mp4, .m4a and .mp3 files.
- Can NotebookLM transcribe audio?Not as a transcript tool. NotebookLM accepts audio recordings as sources and searches their transcribed speech to answer your questions, but Google's announcement describes that use, not a transcript you can export. For a word-for-word copy of the speech, use a transcription tool and then ask NotebookLM about the text.
- Can you dictate or voice type in a Zoom meeting?Not with a Zoom feature. We found no dictation option in Zoom's support pages or its AI whitepaper. You can speak into the Zoom chat box with your computer's own dictation, such as Dictation on a Mac or voice typing on Windows. Zoom's speech-to-text works on the meeting audio, so it makes captions and transcripts, not text you dictate.
- Can you transcribe a Discord call or voice chat?Yes, but we could not confirm a transcription feature built into Discord. You can transcribe a Discord call by adding a bot that joins the voice channel, or by capturing the audio on your computer with a desktop app or a recorder and transcribing the file afterward. Whichever method you use, everyone on the call should know it is being recorded, and some places require every participant's consent.
- Can you transcribe a meeting live, in real time?Yes. Several tools transcribe a meeting as people speak, and some cost nothing: Apple's Live Captions on Mac with Apple silicon and Windows 11 Live Captions both process audio on the device. Live text is a first draft, because words can change as more audio arrives, and Windows 11 does not save its captions. If you need a record, keep the audio and transcribe the final version.
- Can you transcribe a meeting without a bot joining?Yes. You can transcribe a meeting without a bot joining it, using the meeting platform's own transcript, an app that records the audio on your computer, or a microphone in the room. Each needs a setup step before the meeting starts, and each still needs everyone's consent.
- Can you transcribe audio from a link or URL?Some transcription services accept a web address of an audio file, but many need you to upload the file first. Downloading audio from a video site may break that site's terms, so transcribe files you own or have permission to use.
- Can you transcribe audio offline without uploading it?Yes. Open-source Whisper software, and apps built on it, can transcribe a file on your own computer without sending the audio to a server. Download the model while you are online, then test the app with Wi-Fi off before you rely on it for a confidential recording.
- Can you transcribe MP3, M4A, OGG and other audio files?Yes. MP3 and M4A are on OpenAI's list of accepted formats. OGG is not, so convert OGG files to WAV or MP3 before you send them to OpenAI; open-source Whisper hands files to ffmpeg, so OGG may also work there, but test it on a short clip.
- Can you transcribe voice messages from Instagram, Discord and other apps?Some chat apps transcribe the voice messages people send you. WhatsApp, Google Messages and Instagram each include a transcript for messages you receive. WhatsApp's transcripts are off by default and support only a limited set of languages, and Google Messages hides its text until you tap it. We found no official transcript feature in Discord.
- Can you use AI transcription for research interviews?Yes, for most one-to-one research interviews, provided the AI draft is checked line by line against the audio and the consent, data protection and institutional rules are settled before any recording is uploaded. Treat the automatic transcript as a first draft, never as the final record.
- Can you use dictation or voice typing to transcribe a YouTube video?No. Dictation and voice typing take in speech from your microphone and type it where your cursor is, so they are not built to transcribe a YouTube video that is playing on your computer. To get the words, read the transcript YouTube shows, use live captions for the sound on your computer, or transcribe a file you own or have permission to use.
- Can Zoom transcribe a meeting without recording it?Not as a Zoom audio transcript, which Zoom makes only from a cloud recording. You can get live text without recording by turning on automated captions, when the host allows them, and you can save the full transcript as a TXT file while the meeting is still open.
- Does Google Meet transcribe meetings for free?Not with Google's built-in meeting transcripts. Google lists them only for supported Google Workspace editions, and its admin help does not list personal Google accounts among those editions. For a free transcript, you can record the audio on your own device, after checking that recording is allowed, and transcribe the saved file with a free tool.
- How accurate is AI transcription?On clean, single-speaker English, a leading AI model gets close to careful human work. On noisy, distant or overlapping audio, the error rate rises several times, and AI can also invent sentences nobody said, so check any transcript you will rely on.
- How can I transcribe audio to text for free?Yes. The free routes are Apple's Voice Memos on a qualifying Mac, open-source Whisper software that runs on your own computer, and the free allowances that cloud services offer. Use the local route for private recordings and a cloud allowance for a few long ones.
- How can you tell if a transcription job is legit?The fastest test is whether you are asked to pay anything before you have done paid work: a legitimate transcription job never asks for that. Beyond that test, a real platform states its pay per audio minute or hour, its payment method and schedule, and a legal business name you can look up, and it tests your work before paying you. Transcription Staff publishes most of those terms, but reviews report closed accounts and lost pending earnings, and Transcription Hub's public pages show no terms for transcribers, so neither is confirmed as legitimate.
- How do you add automatic captions to YouTube Shorts?Create a checked transcript, time it into short cues, and upload it as an SRT file, or correct any automatic captions YouTube adds to your Short, if it adds any. Automatic text can make errors, especially over music, so review it against the audio before you publish.
- How do you become a transcriptionist with no experience?Start with free practice audio that comes with completed transcripts, then apply to one platform that publishes its terms, and never pay anyone to start. Rev's application tests English grammar and asks for a sample, not a certificate. Formal credentials are a separate route: BLS says medical transcription programs typically take under a year.
- How do you find and download a Zoom transcript?Sign in to the Zoom web portal, click Recordings & Transcripts, open the Cloud recordings tab and click the meeting. The audio transcript is a separate VTT file in the list of recording files, and you can download it and open it in a text editor. It is only there for a cloud recording made with the transcript setting on, and it can take extra time to appear after the recording is ready.
- How do you find freelance transcription work?Apply directly to transcription companies that publish their pay per audio minute and pass a sample test, then choose one or two at a time. Avoid any listing that asks you to pay, and check the pay terms in writing before you start work.
- How do you get a transcript without using AI?Yes. You can listen to the recording and type it yourself, using playback software with a foot pedal or hot keys, or pay a human transcription service to do it. Typing by hand takes about four to six hours for each hour of audio, while a human service does the typing for you for a fee.
- How do you get the transcript of a YouTube video?If the video has captions, YouTube can show you its transcript on the video page, and you can read or copy the text without downloading the video. If it has no transcript, transcribe the audio yourself only if you own the video or have permission to use it, because YouTube's terms bar downloading except where its service permits it.
- How do you make subtitles with Whisper?Run the whisper command on your audio or video with --output_format srt, which writes a numbered, timed subtitle file. Add --word_timestamps True with --max_line_width 42 to break the captions into short lines, then check every cue against the audio, because Whisper's output is a draft.
- How do you mark inaudible speech in a transcript?Put a bracketed tag at the exact point where the words are lost, such as [inaudible 01:28], and add the timestamp so a reviewer can find it. Use one tag set throughout the file, chosen before you start, and tag the gap rather than guess the words.
- How do you prepare interview transcripts for qualitative analysis?Prepare each interview transcript as an editable Word or plain text file, check it against the audio, label every speaker the same way, and replace names with pseudonyms before any coding starts. Edit a copy, and keep the original recording and the first draft unchanged.
- How do you transcribe a lecture recording?Record the lecture with the microphone close to the speaker, run the file through a speech recognition tool, then check the draft against the audio, starting with names, terms and the quiet stretches. A free tool on your own computer is enough for many lectures, but first check that your course allows transcription tools.
- How do you transcribe a long audio file, such as one over 30 minutes?Split a long recording at sentence breaks so each part fits the tool's upload limit, transcribe the parts, then check the joins against the audio. A fast model can process an hour of audio in a few minutes, so the proofreading takes longer than the machine does.
- How do you transcribe a phone call recording?Check first that the other person is willing to be recorded. Apple says both participants hear a notice that the call is being recorded, and call recording on iPhone is not offered in 19 countries and regions, including the European Union. After you have a saved recording, transcribe the audio file with the same tools as any other recording.
- How do you transcribe a podcast?Transcribe the episode with an automatic tool, check every line against the audio, label each speaker by name, and publish the checked text as a WebVTT file if your host accepts one. Apple makes its own automatic transcripts after publication, and its speaker names come from a VTT file that you supply.
- How do you transcribe a QuickTime audio or screen recording?Give the recording to a speech recognition tool that accepts its format. A QuickTime audio recording is usually an AAC file that most tools take as it is. A screen recording is a .mov movie that many tools reject, so export its audio track first, then proofread the draft against the sound.
- How do you transcribe a video to text for free?Three routes cost nothing: copy the transcript the video already shows, transcribe its audio with free open-source software such as Whisper or Buzz on your own computer, or use a free allowance from a cloud service for a few files. Whichever you use, check the draft against the video, and transcribe only a video you own or have permission to use.
- How do you transcribe a Zoom audio or video recording?Find the recording file, then send its audio to a transcription tool. A recording made to your computer has an audio-only file named audio_only.m4a by default, in the Zoom folder in your Documents. Wait until Zoom finishes converting the file, then transcribe it and check names, numbers and dates against the recording.
- How do you transcribe an in-person meeting?Record the room with a microphone close to the speakers, then transcribe the saved file. Distance matters most: a headset or a device on the table near each speaker does far better than a recorder at the far end of the room. Tell everyone the meeting is being recorded before you start.
- How do you transcribe Arabic or Hebrew audio to text?Set the spoken language to Arabic or Hebrew before you transcribe, choose the regional setting the speaker uses, and then check the whole transcript against the audio. On a read-speech test, Whisper large-v2 had a word error rate of 16.0 percent in Arabic and 27.1 percent in Hebrew, so plan to proofread every line.
- How do you transcribe audio in Premiere Pro?To transcribe audio in Premiere Pro, turn on Automatically Transcribe Clips under the Transcription settings, or start a transcript for a clip from the Transcript panel. Then correct the text, check the speakers, and create captions from it. We could not open Adobe's help pages for this review, so confirm the menu names in your version.
- How do you transcribe audio that is playing on your computer?Live captions on a Mac with Apple silicon or a Windows 11 PC show text for the sound playing on the computer, but they do not keep a transcript. To keep the words, record the computer's audio to a file and transcribe that file, then proofread it against the sound.
- How do you transcribe audio with multiple speakers?Use a transcription tool that returns speaker labels, check that it supports your language, and then name each speaker yourself. The tool numbers the voices, but it cannot know who they are, and its labels tend to go wrong where people talk over each other.
- How do you transcribe audio with timestamps for free?Run the open-source Whisper command-line tool on your own computer, which is free to use. Add the --word_timestamps True flag for word-level timestamps, an option the source marks as experimental, and check the times against the audio, because Whisper's own timestamps can be inaccurate by several seconds.
- How do you transcribe Hindi or Hinglish audio to text?Yes. For a recording, set the language to Hindi and transcribe it with Whisper or an app built on it, such as Buzz, which runs on your computer. Hindi is recognized less accurately than English, and Hinglish, which mixes English words into Hindi, is harder still, so check the transcript against the audio before you use it.
- How do you transcribe Japanese, Chinese or Korean audio?Yes. Whisper's language table includes Japanese, Chinese and Korean, and Google Cloud's speech service lists Japanese too. Set the language and, for Chinese, the Simplified or Traditional script, then check the transcript by reading it, because Chinese and Japanese have no spaces between words.
- How do you transcribe low-quality or quiet audio?Start with the recording: move the microphone close, reduce background noise, and transcribe the original file rather than a compressed copy. Then raise the level with a normalizing filter and proofread the quiet parts, because speech recognition can invent words where there is little speech.
- How do you transcribe Spanish, German or Dutch audio to text?Set the spoken language to Spanish, German or Dutch yourself, transcribe as usual, and check the result against the audio. On a read-speech test, Whisper large-v2 had a word error rate of 3.0 percent in Spanish, 4.5 percent in German and 6.7 percent in Dutch. Regional accents and dialects still change the result, so check the transcript before you use it.
- How do you transcribe Tamil, Urdu, Bengali or other South Asian audio?Set the spoken language to Tamil, Urdu or Bengali yourself, transcribe with a model that supports it, and check every line. On a read-speech test, Whisper large-v2 had a word error rate of 17.5 percent in Tamil and 22.6 percent in Urdu, but 104.1 percent in Bengali, so on that model the Bengali output is not a usable draft. The same steps apply to other South Asian languages, with each one checked on its own.
- How do you translate a voice recording into English text?The reliable way is two steps: transcribe the recording in the language it was spoken in, check that transcript against the audio, then translate it into English. A few tools do both in one step, such as Whisper's translate task, but recognition errors carry into the English, so check the transcript either way.
- How do you turn a meeting transcript into minutes?Turn a transcript into minutes by writing down only what was decided: each motion with its exact wording, who moved it, the result, and each action with an owner and a deadline. Check every entry against the audio. Minutes record what was done, not what was said, so do not paste the transcript or an AI summary in as the minutes.
- How do you turn a voice recording into text on Android?Start with the recorder on your phone. Pixel Recorder has Live Transcription and a Transcribe again option, and Samsung Voice Recorder transcribes a saved recording with Transcript assist on supported Galaxy models. For a recording with no built-in transcript, export the audio file and transcribe it with a file tool.
- How do you turn off or disable transcription in Zoom?Clear Create audio transcript under Advanced Cloud Recording settings in your Zoom account settings. An account owner or admin does this for a business account, and a personal account owner can do it too unless an admin has locked the setting. You cannot change it during a meeting, and it applies only to cloud recordings. To stop live captions in a meeting, the host turns them off from the captions controls.
- How do you turn on transcription in Zoom?Turn on Create audio transcript in Zoom's recording settings. Zoom makes a transcript only from cloud recordings, so the meeting must be recorded to the cloud on a Pro, Business, Education or Enterprise account, and the setting must be on before the meeting starts.
- How do you use Whisper to transcribe audio?Install the openai-whisper package and the ffmpeg tool, then run one command on your audio file, such as whisper recording.mp3 --model turbo. Whisper is free and open source, and the openai-whisper package runs on your own computer, so the recording does not have to be uploaded.
- How long does it take to transcribe one hour of audio?An AI engine transcribes one hour of audio in seconds to a few minutes, depending on the engine. Transcribing it by hand takes about four to nine hours, and Rev, a human service, promises the text within 12 hours, so the time you need depends on whether you need the text now or a finished file.
- How much do transcriptionists make?A general transcriptionist is paid per audio minute, so a beginner earns less per hour than the posted rate once listening, replaying and checking are counted. At Rev's published range, one audio hour pays $24 to $66, and a typist at average speed earns $8.23 to $22.63 for each hour of typing alone.
- How much does transcription cost per minute?AI transcription APIs list at about half a cent to six-tenths of a cent per minute of audio, under a dollar an hour. Rev, a human service, lists $1.99 per minute, roughly 440 times the AI price, so an AI draft with a human check of the hard parts starts far below the price of a full human service.
- How often should you add timestamps to a transcript?There is no single rule. Oregon State University's oral history guidance uses timestamps at five-minute intervals, which helps a reader find a passage in a long interview. Add a timestamp at each speaker change when you need to find who said something, and use word-level timing when you need exact positions.
- Is there a free meeting transcription app?Yes, for short or occasional meetings. Otter's free Basic plan allows 300 transcription minutes a month, with a 30-minute cap on each conversation, and open-source Whisper tools are free to run on your own computer. A platform transcript is free only when your plan or organization includes it.
- Is transcription a good side hustle?Transcription can work as a side hustle if you already type fast and accurately and can fit it around other work. It is a weak choice if you need a high hourly rate, because it is paid by the audio minute and typing time can push your real hourly pay well below the posted rate.
- Is transcription software considered generative AI?Transcription software is AI, because its speech recognition is built on machine learning. It is generative AI only when a language model writes new text after recognition, such as a rewrite, a cleanup, a summary or an answer about the audio. A plain transcript sits at the edge of the definition, so check what each app does with the words it hears.
- Is transcription software HIPAA compliant, and how do you check?No transcription software is HIPAA compliant on its own, because HIPAA judges how an organization uses a tool with protected health information, not the product's name. To check a tool, confirm three things: the vendor will sign a business associate agreement that covers that exact product, the vendor's logging and retention settings match that agreement, and your own HIPAA risk analysis covers the tool.
- Is transcription still in demand now that AI exists?Demand for medical transcriptionists is projected to fall, and more of the remaining work is reviewing and correcting machine drafts. The US Bureau of Labor Statistics projects medical transcriptionist employment to decline 4 percent from 2025 to 2035, with openings coming from replacing workers who leave, and it expects AI to dampen demand for court reporters too.
- What does a good transcript look like?A good transcript labels every speaker, marks time so a reader can find any passage in the audio, keeps the words in the order they were spoken, and marks what cannot be heard. Choose the style, full verbatim or clean verbatim, before you start, because it decides which words stay on the page.
- What does it mean to transcribe audio?To transcribe audio means to listen to a recording and write down what is said, word for word, as text. Software can now produce a usable first draft quickly, but someone should proofread that draft against the sound before you rely on it.
- What equipment do you need for transcription work?You need a reliable computer, closed headphones that you can keep at a safe volume, and transcription playback software. A foot pedal is useful but optional, because the software can run playback from the keyboard, and the employer's requirements set what you must buy.
- What is an SRT file, and how is it different from VTT?An SRT file is plain text of numbered entries, each with a start time, an end time and the text, and its timestamps use a comma before the milliseconds. A VTT (WebVTT) file is the web version of the same idea: it starts with the word WEBVTT and uses a full stop before the milliseconds. Use SRT for uploads and desktop tools, and WebVTT for web players.
- What is Express Scribe and what are the alternatives?Express Scribe is a free audio player from NCH Software that plays a recording so you can type it, controlled by keyboard hot keys or a foot pedal. It does not transcribe speech by itself. Alternatives include oTranscribe, a free browser player, and automatic services when you do not need to type each word.
- What is the best open-source meeting transcription tool?There is no single best tool. For most people, the starting point is a tool built on OpenAI's Whisper, which is MIT licensed. Meetily is a meeting app that captures system audio and lets you choose its summary provider. For speaker labels, use WhisperX, and for live text, expect a Whisper stream with lag. Check the license of each model you download, not only the app.
- What is the difference between a meeting transcript and meeting minutes?A meeting transcript is a record of what was said, taken from the audio. Meeting minutes are the official record of what was decided and done: motions, votes, outcomes and actions. The transcript is the raw material for the minutes, not a substitute for them, so write the minutes from the transcript and check each entry against the audio.
- What is the difference between captions and subtitles?Captions are timed text for viewers who cannot hear the audio, so they include speaker names and sounds. Subtitles are timed text for viewers who can hear but do not understand the language, so they carry the dialogue, often translated. Choose by the viewer's problem: if they cannot hear, make captions; if they hear but need the language, make subtitles.
- What is transcription QA, and how do you proofread a transcript?Transcription QA (quality assurance) is the check that a finished transcript matches its audio and its style rules. To proofread it, listen to the audio while you read, focus on speaker names, numbers and jargon, and use a second person for high-risk work.
- What typing speed do you need for transcription?Rev, the platform we checked, publishes no words-per-minute minimum. It lists "fast, accurate typing" among the skills for its transcriptionist role and asks for a sample that must meet its quality standards, so accuracy counts as much as speed. For a benchmark, the average online typist managed 51.56 words per minute, and the fastest tenth typed above about 78, while court reporting tests set bars of 180 to 225 words per minute.
- Why is auto transcription not working in CapCut, Canva or Premiere Pro?CapCut's own help article lists the causes: poor or missing audio, a wrong language setting, a weak connection and an outdated app. In Premiere Pro, check that Automatically Transcribe Clips is on and that your language is supported. We could not open Canva's or Adobe's help pages, so we give no product steps for Canva.
- Why is the YouTube transcript not showing?Most likely the video has no transcript because it has no captions. Check for a CC button, then the caption language, before you assume a fault. We could not open YouTube's help page to confirm this list, so treat it as a set of checks rather than a guarantee.
- Why is voice memo or voice message transcription not working?A voice memo or voice message transcript can fail because of the setup: the phone is too old, the language or region is not supported, the transcript switch is off in a chat app, or the recording was deleted. Check those in that order. If they all pass, the audio may be the cause, so transcribe an export with a file tool and check the draft against the sound.
Terms in this topic
- Clean verbatimClean verbatim is a transcript that removes filler words, stutters, repetitions and false starts while keeping the speaker's wording and meaning. It is the usual choice for reading, captions and coding, where hesitations add noise rather than information.
- Closed captions (CC)Closed captions (CC) are a text version of a program's dialogue and relevant sounds that the viewer can turn on or off. The word closed means the viewer chooses whether to see them, which separates them from open captions.
- Forced alignmentForced alignment is the process of matching a known transcript to its audio so that each word or phone gets a start and end time. Unlike open speech recognition, it does not decide what was said; it only finds when each word was said.
- Open captionsOpen captions are text drawn onto the frames of a video, so every viewer sees them and nobody can switch them off. They are also called burned-in captions or hardcoded subtitles.
- SubRip subtitle file (SRT)A SubRip subtitle file, or SRT, is a plain-text file of numbered caption blocks. Each block has a start and end time in hours, minutes, seconds and milliseconds, followed by the caption text. It is one of the most common caption formats on the internet.
- Subtitles for the deaf and hard of hearing (SDH)SDH stands for subtitles for the deaf and hard of hearing. They show the dialogue plus the sound effects, music and speaker identification that a hearing viewer would get from the soundtrack.
- Timestamp (transcription)A timestamp in a transcript is a time marker, such as 00:12:15, that shows when the nearby text was spoken in the recording. Timestamps let readers jump to a passage, and they can be attached to whole segments or to single words, with different accuracy.
- Transcription foot pedalA transcription foot pedal is a switch that controls audio playback with the foot, so a transcriber can start, pause and rewind a recording while typing. It is also called a foot control or foot pedal.
- TranscriptionistA transcriptionist is a person who listens to recorded speech and types it as text, such as a clinician's dictated report. Speech recognition now writes first drafts, so much of the work has shifted to checking and correcting machine text.
- Turnaround time (TAT)Turnaround time (TAT) is how long a transcriber or service takes to deliver a finished transcript after an order is received. Court schedules define it in calendar days or hours, counted from the order.
- Verbatim transcriptionVerbatim transcription writes down exactly what is said, including filler words, false starts, repetitions and non-speech sounds, with no tidying. Researchers and legal teams use it when the way someone speaks is part of the record.
- WebVTT (VTT)WebVTT (Web Video Text Tracks, usually shortened to VTT) is a plain-text file format for timed captions and subtitles on the web. A .vtt file is a list of cues, and each cue gives a start time, an end time and the text to show.
Referenced by
- Dictation and Voice Typing: How to Talk Instead of TypeTopic
- Dictation Software: How to Choose the Right AppTopic
- Speech to Text for Accessibility, RSI and DisabilitiesTopic
- Dictation for Work: Doctors, Lawyers, Writers, StudentsTopic
- Dictation and Voice to Text on iPhone and AndroidTopic
- Speech Recognition: What It Is and How Well It WorksTopic
- Speech to Text Privacy: Where Your Voice GoesTopic
- Dictation Languages, Accents and Voice TranslationTopic
- Speech to Text AI: Dictation, Assistants and Voice AgentsTopic
- Dictate in Word: Voice Typing on Web, Windows and MacGuide
- Meeting Transcription: How It Works and What to UseGuide
- Zoom, Teams and Google Meet TranscriptionGuide
- Video to Text: Transcribe Videos and YouTubeGuide
- Voice Memo to Text: Built-In and Free OptionsGuide
- Transcription Software: How to Choose OneGuide
- Human vs AI Transcription: Accuracy, Cost, SpeedGuide
- Legal Transcription: The Job, Formats and PayGuide
- Transcribe Audio With Whisper, ChatGPT or GeminiGuide
- Transcription Jobs: Pay, Skills and Getting StartedGuide
- Captions and Subtitles: Formats, Rules and ToolsGuide
- Dictation for Academics, Researchers and ThesesGuide
- Verbatim vs Clean Verbatim TranscriptionGuide
- Timestamps and Speaker Labels in TranscriptsGuide
- Transcribe and Translate Audio in Other LanguagesGuide