One of the strongest languages
After English, Spanish is among the languages speech models handle best, simply because there is so much Spanish audio in the world. With the Standard and High accuracy models it is rated high: clear Spanish speech comes back with few mistakes, correct accents on words such as “está” and “también”, and punctuation, including the opening ¿ and ¡ marks. Fast is rated medium for Spanish; it is fine for a quick look but leaves more to correct.
Spain, Mexico, Argentina and everywhere else
The model was trained on Spanish from many countries, so it follows Castilian, Mexican, Caribbean, Andean and Rioplatense speech. Vocabulary that differs between countries, such as “coche” and “carro”, comes out as it was said. Fast regional speech, dropped final consonants in Caribbean Spanish, and the “sh” sound of Buenos Aires can cause occasional slips, so the High accuracy model is the better choice for strongly regional recordings.
Mixed Spanish and English, common in the United States, is harder. The engine follows one language per file; keep Spanish selected and fix English phrases by hand, or transcribe in English if most of the recording is English.
Choose Spanish rather than detect
Language detection listens to the start of the file. A recording that opens with English, music or silence can be guessed wrong, and the whole transcript then comes out in the wrong language. Coming from this page, Spanish is already selected, and it stays your choice in the workspace until you change it.
Spanish to English
Spanish is also one of the languages that translates best into English. With a multilingual Standard or High accuracy model, tick Translate to English before you start and the engine writes English text directly from the Spanish speech, timed for subtitles. For an exact record, transcribe in Spanish too and keep both.
Checking the result
Look first at names of people and places, which are often spelled the way they sound. Check numbers and dates, and words that differ only by an accent, such as “sé” and “se” or “más” and “mas”, where the model sometimes picks the wrong one. Click any line to hear it again while you correct it.
Dictating in Spanish
Spanish also works well for live dictation. Press Record, speak at your usual pace, and the live draft follows you; with the Standard model it is already quite readable. When you stop, the final pass fixes most of what the draft missed.

