QQuestions
Q.01How do I translate an audio file?
Upload the file above, pick the language spoken in it and the language you want, tick the three consent boxes and press Start translating. We transcribe it, translate it, show you the translation to correct, then generate the new audio track. A 10-minute file usually takes a few minutes.
Q.02Can I translate an audio file to English?
Yes. Choose Spanish, German, French, Japanese or Chinese as the source and English as the target. See translate audio to English for tips.
Q.03Will the translated audio really sound like me?
If you choose Keep my own voice, we clone your voice from a clean 10–60 second stretch of your own recording and use it to read the translation. It is close, not perfect: accents and some expressiveness shift in another language. You can always switch to a standard voice instead.
Q.04Do you translate video or do lip-sync?
No. We only replace the audio track: you get a translated MP3. If you have a video, export its audio, translate it here, and put the new track back in your editor. We do not change mouths or faces.
Q.05Can I upload someone else's podcast or audio?
Only if you own the rights or have written permission, and only clone a voice that is yours or that you have explicit consent to use. Every upload records your confirmation.
Q.06How long do you keep my files?
Uploads, transcripts and dubs are deleted automatically 7 days after upload, and you can delete a job instantly from its page. Your audio is never used to train models. Details in our privacy policy.
Q.07Which languages can I translate audio between?
English, Spanish, German, French, Japanese and Chinese (Mandarin), in any direction — 30 pairs in total.
Q.08Which file formats are supported?
MP3, M4A, WAV, AAC, FLAC, OGG and Opus, plus the audio inside MP4/WebM files. Maximum 95 MB and 120 minutes per file.
Q.09How do the number and date checks work?
After translation, every number in a source sentence must appear in its translation (we accept 1,000 / 1.000 style differences). Lines that fail are flagged in the review step so you can fix them before voicing.
Q.10What does “timeline-preserving” mean?
Each translated sentence starts at the same timecode as the original sentence, and pauses between sentences are kept close to the original. Speed is adjusted within 0.85–1.15× to fit; very long translations may still shift slightly.
Q.11Is the output labelled as AI?
Yes. The MP3 metadata states that the audio is AI-generated, the download file name says so, and the job page shows the label. Please keep a clear AI-voice disclosure when you publish.
Q.12Do you use my audio to train AI?
No. Your audio, transcript and voice clone are only used to produce your job.
Q.13What about multiple speakers?
The current version treats the file as one main speaker. Every person who speaks must consent to their voice being used; multi-voice dubbing is planned.