Spanish Speech to Text and Talk to Text

To dictate in Spanish, choose the Spanish variant you speak in your device's dictation or voice typing settings, then speak at a normal pace. Spanish is one of the languages that speech recognition handles well. In a published benchmark, Whisper large-v2 had a word error rate of 3.0 percent on Spanish read speech ¹, against 4.2 percent on English ². Two things need your attention. The first is the regional variant you select. The second is the written accent, because you never type it, so the recognizer has to choose it for you.

Spanish is supported on the major platforms, and recognizers handle it well

Spanish is available for voice typing on the main desktop and office tools. Windows 11 voice typing lists two Spanish variants, Spanish (Mexico) and Spanish (Spain) ³. Microsoft's Dictate in Word lists 15 main languages, which include both Spanish (Spain) and Spanish (Mexico) ⁴. Windows Voice Access, the hands-free control feature, lists Spanish among six languages on Microsoft's command list ⁵. Microsoft's pages do not agree on that count, so check the setup page for your variant.

On a Mac, Dictation draws from 63 language and regional variants ⁸, and Apple's feature list names five Spanish variants for on-device dictation: Chile, Colombia, Mexico, Spain and the United States ⁹. Open Keyboard settings to see which Spanish entries your Mac offers, because the list depends on your macOS version. Our overview of dictation on a Mac explains the steps, and Dictation Languages compares the lists across devices.

Adding a Spanish variant on Windows 11 takes four steps:

1. Select Start, then Settings.
2. Select Time & language, then Language & region.
3. Under Preferred languages, select Add a language, and search for the Spanish variant you want.
4. Select Next, then install any optional features you want.

Microsoft says a voice typing language can differ from the language you chose for Windows ¹⁰, so you can keep Windows in English and dictate in Spanish.

Pick the regional Spanish you speak: Spain, Mexico, the United States and more

Choose the variant that matches your own speech, and expect the platforms to list different sets. Windows 11 voice typing and Word both offer Spanish (Mexico) and Spanish (Spain), so those are the variants the sources confirm for both tools. A variant is a language and region setting, and it tells the recognizer which pronunciation and vocabulary to expect. Our guide to dictation languages compares the language lists across vendors.

The difference between variants is real in speech as well as in menus. Most of Spain separates the sound of the letters z and c before e and i from the sound of s, while most of Latin America does not, and the everyday vocabulary also differs. A computer is usually called a computadora in Mexico and an ordenador in Spain. A variant that matches your vocabulary gives the recognizer better odds when two words sound alike. These are general features of the two varieties, not measured dictation results, so the test below is how you confirm the choice for your own voice.

Accuracy by variant is harder to check than accuracy by language. The sources we reviewed do not publish dictation error rates for each Spanish variant, so no variant can be ranked for you from published figures. Test the variant you choose by dictating a paragraph of your own writing, then compare the result with what you said. Correct the words the recognizer gets wrong, and if the variant fails often, try the other one. The question of which Spanish to choose for dictation gives the same test in short form.

Command phrases can also differ between variants, so check the command list for yours before you rely on it.

Here is a quick test you can run in about ten minutes:

1. Open the app you dictate with most, and select the Spanish variant you speak.
2. Read aloud a 100-word passage you wrote yourself. Include a question, a name, an exclamation and several words with accents, such as está, qué and según.
3. Count the words the recognizer got wrong, and note each missing tilde, missing ¿ or wrong punctuation mark.
4. Repeat the test with the other variant if the first one made many errors.
5. Keep the variant with fewer corrections, and add the names you use often to the app's custom dictionary if it has one.

Accent marks and the ñ come from the recognizer, so you never type them

When you dictate, you do not type accent marks, so the recognizer must write them for you. The ñ is a separate letter in Spanish, and it is recognized like any other letter. Tildes on vowels must be decided from how the word sounds, and that decision can go wrong. A 2016 study from the Universidad de Costa Rica describes the omitted tilde as one of the most common errors of automatic transcription in Spanish ¹¹. The authors blame mainly the techniques that ignore stress, which is the syllable a word puts its emphasis on ¹². Their paper proposes using intonation to decide where the tilde belongs, so it describes a known weakness and does not measure how often it happens today.

Spanish has many pairs of words that sound the same and differ only by an accent. Check these in every draft:

If a tilde is missing, type it after dictation. Keep the review step even on a good result, because a missing mark can pass unnoticed in a long paragraph. Read the Spanish once aloud, or check a few sentences with a question or a name, which are the places a recognizer is most likely to slip.

These homophones are the cases where a missing accent changes the meaning, so they deserve a quick look after dictation.

Spanish also has command words that are everyday words. Punto means a period in a dictation command, but it is also the ordinary word for a point or a dot, and coma is both the comma mark and a common word for a comma-shaped stroke or a medical term. Say each command clearly and check that the app inserted a mark rather than the word. Commands that fail in a noisy room are often the ones that sound like ordinary speech. The question of whether dictation adds accent marks has the same answer: it depends on the product and the word, so check the output. The diacritics entry explains what the marks are called and where they appear.

Spanish had a lower error rate than English on the Fleurs benchmark

On the Fleurs benchmark, which uses read speech, the large-v2 model reached 3.0 percent word error rate for Spanish ¹ and 4.2 percent for English ². That makes the English error rate about 1.4 times the Spanish rate ¹³. On the same benchmark, Spanish also scores better than German, French and Hindi, as the chart shows.

The chart uses the figures from the Whisper paper: Spanish ¹, English ², German ¹⁴, French ¹⁵ and Hindi ¹⁶. Hindi is included to show how much the error rate can rise from one language to another.

These figures have limits, and the table shows where your dictation differs from the benchmark.

The paper tests one model, not the built-in dictation of any particular device. The sources we checked do not publish a Spanish error rate for Apple or Microsoft dictation, so we cannot say how they compare. For a fuller explanation of the measure, see the word error rate entry and our question on whether Spanish speech to text is accurate.

Accent and dialect change the result too. A 2026 preprint on YouTube's automatic Spanish captions, which is not a dictation test, found systematic differences in how accurately the captions handled different dialects ¹⁷. Our guide on speech recognition and accents covers the fixes. The accent and dialect entries define the terms.

Spanish punctuation is spoken in Spanish and includes opening marks

Spanish puts an opening mark before a question, so a question has both ¿ and ? ¹⁸. A dictated Spanish sentence needs those marks too, and the marks you need depend on which command phrases your app knows. In Microsoft's Dictate in Word, you say the phrases below to insert each mark ¹⁹.

Automatic punctuation is a separate feature. Apple describes it as adding commas, periods and question marks automatically in supported languages ²⁰, and it lists four Spanish variants, not including Spanish (Chile) ²¹. That description does not mention opening marks, so do not count on ¿ or ¡ appearing on their own. In Word, say the phrase. On a Mac, Apple's Spanish command table has no spoken command for ¿ ²², so type the mark after dictation. The automatic punctuation entry explains how the feature works in general, and our guide on dictation punctuation in other languages compares the commands for French, German and Spanish.

Bilingual speakers can switch or translate between Spanish and English

Dictation writes the language you select. On a Mac, you can set up several dictation languages and move between them, using the Globe key or the language label next to the cursor ²³. Apple describes switching between languages as needed ²⁴, and its page does not say how Dictation handles one sentence that mixes two languages. If you write in both languages within one sentence, see our guide to multilingual dictation.

The iPhone keyboard offers a bilingual option for some pairs, and its example pairs English (U.S.) with Spanish (Latin America) ⁷. That lets you type in both languages without changing keyboards. Apple describes the option for typing, not for Dictation.

You have three practical routes to English. The table compares them.

The first route takes two steps. First, dictate or transcribe the Spanish. Second, translate the text. For the second route, OpenAI's speech to text guide names whisper-1 for translating a completed recording into English ²⁵. The turbo Whisper model is not trained for translation, so translating into English needs a multilingual model such as medium or large ²⁶. Errors add up across the two steps, so read the English output against the Spanish. Our guide on speech to text with translation covers the tools, and the question on talking to text in Spanish and getting English gives the short version.

OpenAI says Whisper supports 98 languages, but accuracy varies by language ²⁷, so measure Spanish on your own audio. Whisper is a model, not a dictation app. If you want to test Spanish speech in a browser without installing anything, online dictation runs in the browser, and audio to text transcribes a recording you already have.

Next steps for Spanish dictation

Start with the variant the platform offers closest to your speech, then run the test above. Make the decisions in this order: the platform, the variant, the accent check, and last the translation route if you need English. Each step depends on the one before it, so a wrong variant makes the later checks harder to interpret. Dictate a paragraph and check the accent marks, the ñ and the opening question mark. If the output is clean, you can rely on it for drafts. If accent marks go missing, keep them in your review step. The question on whether Spanish dictation adds accent marks and the Mac and Windows pages list the settings for each system.

If you want a dictation app that runs on your own Mac or Windows PC, Yap, which we make, is one option. Compare it with the built-in tools above before you choose, and see the download page for what it supports today. Our overview of dictation languages covers the wider set of languages, and the languages overview links to every guide in the series.