Podcast Translator

Why it helps

Podcast audiences are fragmented by language, and new-language feeds are a proven way to grow — but hiring voice actors for every episode is expensive. AI translation makes a pilot feed affordable.

Step by step

1. Export the episode as MP3 or WAV.
2. Choose the target language and a voice; a calm or narrator voice suits interview shows.
3. Download the translated M4A — your intro music and stings stay in the mix — and publish to a separate feed.

Tips

• Translate a back-catalogue favourite first and share it with a few native-speaking listeners for feedback.
• Tell listeners the episode is AI-translated — transparency builds trust.

Example

Cost: 2,050 credits ($2.05).

Limitations to know

• One voice speaks for everyone; multi-host shows lose the distinction between hosts.
• Music is kept under the translated voice, but separation isn't perfect: sung vocals in music are removed, and faint traces of the original voice can remain where music is loud.

Decide what the translated show will be

Separate feed

A new show in the target language with its own title, artwork and description. Clearest for listeners, and directories list it as a show in its own right.

Episodes in the main feed

Translated episodes added to your existing feed, with the language in the title. Quick to launch, but confusing for subscribers who don't speak that language.

Translated transcripts only

No audio: a transcript-mode run with a target language at 1 credit per minute, posted on the episode page. A low-cost way to test interest.

Shows that commit usually go for a separate feed; multilingual podcasts compares the setups in more depth.

Preparing the source audio

What you upload shapes what comes back. If you upload the full mix, its music beds and stingers are kept under the translated voice, though loud music under speech makes both recognition and separation harder. If your editor can export the conversation without the music, recognition has less to untangle, but you will then add the music to the translated episode yourself. Cut long cold opens built from clips of later moments, since they will be translated again when the clip appears in full. And if the episode is close to two hours, split it at a natural break, such as the halfway sponsor slot, and join the translated halves in your editor.

Finishing the episode around the translated voice

The translated M4A is the translated voice with the original background mixed underneath, in mono. It is close to a finished episode, but a few steps remain.

1. Import the M4A into your editor on its own track.
2. Check the intro, outro and transition music. Sung lyrics are removed along with the speech, and in dense mixes traces of the original voice can remain; if you have clean music stems, dubbing a dialogue-only export and mixing the translated voice with them gives the cleanest result.
3. Deal with sponsor reads: re-record them with a speaker of that language or drop them, depending on your agreement. See sponsored segments in translated content.
4. Normalise loudness to the same target as your main feed so listeners don't reach for the volume; loudness normalisation explains the targets.
5. Export at your usual podcast settings and upload to the new feed.

A short spoken intro from the host, recorded in the new language if they can manage a few lines, makes the translated show feel less anonymous.

Interview and panel shows

One voice reads the whole translated episode. In a solo or narrated show that's barely noticeable; in an interview it means host and guest sound identical, so listeners rely on the content to know who is talking. A few things help: keep questions clearly phrased as questions in the edit, add a short music sting between long answers, and introduce the guest in the episode description. If the format depends on banter between several hosts, listen to a full translated episode before committing; single-voice vs multi-voice dubbing discusses when one voice works. Pick the voice with choosing an AI voice in mind, and keep it the same across episodes.

Budget for a pilot season

Give the pilot a fixed run of several episodes and judge it on downloads and messages from listeners in that language, not on the first week. New feeds start with no subscribers, and directories take time to surface them, so an early quiet spell says little about long-term interest.

Telling listeners and gathering feedback

Say at the start of each translated episode, and in the show description, that the voice is AI-generated from the original episode. Listeners are more forgiving of a stock voice when they know what they are getting. Ask a handful of native-speaking listeners to report anything that sounds wrong, such as names, numbers or idioms translated literally, and keep a list. Fixing a single word in finished audio is fiddly, so those notes mostly improve how you prepare the next episode: spelling out an unusual name on air, or avoiding a pun that won't survive translation. The audio translator page lists every download, and podcast transcription covers transcripts in the original language.

Using the tool on this page

• The spoken language is detected automatically; you select the output language — the spoken language itself for an untranslated transcript, or another language for a translation. Files can be up to 2 hours long.
• Audio is sent over HTTPS in small chunks and deleted with its transcripts within 30 minutes of delivery (how files are protected).
• You can download the translated audio (voice over your original music and effects) as M4A, a transcript and subtitles (SRT and VTT). Every output, step and limit is explained on the audio translator page.