What we tested

On 10 November 2026 we cut 59.5 seconds of English dialogue between a man and a woman from Tears of Steel, the Blender Foundation's open film, saved it as an MP3 and asked Vocnu for a Spanish voice-over as audio, in Premium.

  • It cost 2 credits: under a minute is billed as one, at 2 credits a minute in Premium.
  • Back came an MP3 of 59.533 seconds, exactly as long as the upload, at 48 kHz stereo and 192 kb/s.
  • With it, the Spanish subtitles as SRT, VTT and ASS files, and the English transcript as an SRT file.
  • One male voice read every line, hers too. We measured the pitch: her line is around 296 Hz in the original, and every line of the Spanish file sits between 107 and 123 Hz.

An earlier cut of exactly 60 seconds came out of our MP3 encoder at 60.029 seconds and was quoted 4 credits, two started minutes. A clip close to a whole minute is worth trimming a little.

Files in, files out

  • In: MP3, M4A, WAV, AAC, OGG, Opus, FLAC and other common audio and video formats, up to 10 GB and 5 hours.
  • Out: the voice-over as M4A, MP3, WAV, FLAC or OGG, as long as the episode.
  • Your original sound stays under the voice, lowered while it speaks, so listeners still hear the music and the hosts' laughter. Or turn it off, and the file holds the voice and silence between its lines.
  • The subtitles in the new language as SRT, VTT and ASS files, and a transcript of the original, for show notes or a video version of the episode.
  • One job makes one language. For the next, start a new job from the same upload within 7 days, without uploading it again.

What an episode costs

QualityAn hour-long episodeHow it sounds
Premium€7.92 to €12Our most natural voices and finest translation
Standard€3.96 to €6.00Clear voices and a good translation
Economy€1.98 to €3.00A plain synthetic voice and a quick translation

The prices run from the largest credit pack to the smallest, and the credits never expire. You see the quote before the job starts and are never charged more; failed or cancelled jobs are refunded in full.

When Vocnu is not the right pick

  • Your show is an interview or a panel, and listeners must tell the voices apart. Vocnu reads every speaker in one voice, as our test showed; ElevenLabs keeps each speaker's voice.
  • You want the episode in your own voice. Vocnu reads with its own voices; ElevenLabs and Rask AI can clone yours.
  • You record live and want it translated as you speak. Vocnu translates a finished file.
  • You want to check and correct the translation line by line before it is voiced. Vocnu has no editor for that.

Publishing a video version on YouTube? See Vocnu for YouTube creators.

Questions and answers

Can listeners still hear my voice?

Yes, lowered under the translation, unless you turn the original sound off. Then only the new voice remains.

Can each guest have a different voice?

No. The voice you choose reads every speaker.

Why did a 60-second clip cost two minutes?

Minutes are rounded up for each file, and an MP3 encoder can add a few milliseconds: our exactly 60-second cut came out at 60.029 seconds, so it was billed as two minutes. Trim a short clip a little below the minute.

Can I translate someone else's podcast?

Only if you own it or are allowed to translate and adapt it, as our acceptable use policy says. Having a copy of an episode does not give you that right.

Sources

Facts checked on November 10, 2026.

Try it on your own video

Sign up, get 40 free credits and see the price for your video before you spend anything.

Create free account