01Spanish for listening, not for reading
A written translation and a spoken one are different jobs. We ask the translation model for Spanish that sounds right out loud: shorter clauses, everyday vocabulary, and a length close to the original sentence so the Spanish fits the same moment on the timeline.
Spanish typically runs 15–25% longer than English. To stay in sync we give every sentence a character budget based on how long you spoke it, and the voice can speed up a little where needed.
02Which Spanish?
By default you get neutral Latin American-leaning Spanish that works for US Hispanic audiences and Mexico. If you publish for Spain, use the review step to adjust words like computadora/ordenador or ustedes/vosotros; the voice will follow your edits.
03Check names, numbers and dates
Before anything is voiced, each sentence sits next to its English source. If a figure from the English — a price, a percentage, a year — is missing from the Spanish, the line is flagged so you can fix it. Product names stay as you said them unless there is a standard Spanish form.
04What you get
- A Spanish MP3 on the same timeline as your English original, loudness-normalised
- A side-by-side English / Spanish transcript with timecodes
- AI-generated labelling in the file metadata, ready for disclosure when you publish
05Good to know
Clear speech with one main speaker gives the best results. Music beds and crosstalk lower transcription quality. Uploads up to 120 minutes and 95 MB; files are deleted after 7 days.
QQuestions
Q.01Can I translate Spanish audio to English instead?
Yes — set the source to Spanish and the target to English, or see translate audio to English.
Q.02Will my voice speak Spanish with an accent?
Some. The clone keeps your timbre; the Spanish pronunciation comes from the speech model and is generally clear, but it won't be identical to a native speaker.