Samevoice

Your video in another language, in the same voices.

Samevoice dubs interviews, podcasts and lectures between Russian and English and between German and English. Each speaker is voiced by a clone of their own voice, and a professional translator checks every dub in our browser editor before you get it.

Illustration: an astronaut wearing headphones, with a sound wave passing from one speech bubble to another.

Example: 15 seconds of a two-person interview

Dub into
  1. 1Host

    You've said the hardest part was the first year.

    Вы говорили, что труднее всего был первый год.

  2. 2Guest

    It was. Nobody tells you how quiet it gets.

    Так и было. Никто не предупреждает, какая наступает тишина.

  3. 3Guest

    And then one day it isn't quiet anymore.

    А потом однажды тишина заканчивается.

Each dubbed line starts where the original line starts and is voiced in that speaker's own voice. The dashed clip is the host's laugh. It and the music are the original recordings, kept when you send separate tracks.

How a video gets dubbed

Software does the transcription, translation and voicing. A person decides when the dub is ready.

  1. You send the video

    A YouTube link is enough. For the best result, send your editor's project with one audio track per speaker and a separate track for music and effects.

  2. It is transcribed and translated

    Speech is transcribed with the timing of every word, split into sentences and translated with the time each line has to fit in mind.

  3. Each speaker is voiced in their own voice

    A short sample of each person's speech from the video becomes the reference for their dubbed lines. Every line is then transcribed back and compared with its script, and a line that doesn't match is voiced again.

  4. A translator reviews the dub

    A professional translator listens to the result in the editor, corrects wording and pronunciation, and chooses a better take where a line needs one.

  5. You get the dubbed video

    The new voices are mixed at the loudness of the lines they replace and joined to your picture, which is left untouched.

The editor

Every dub opens in a browser editor, where the video, the original lines and the dubbed lines sit side by side. Our translators work in it, and your own reviewer can have an account too.

An illustration of the editor, using the example lines from the top of the page.
  • The original and the dub on one clock

    Switch between them at any moment of the video. The lines follow the playhead, and a click on a line jumps to it.

  • The lines that need a look come first

    One filter gathers every line the automatic checks flagged, so a reviewer starts where the problems are.

  • Change a line and hear it again

    Rewrite the translation, mark the stress in a Russian word, leave a line unvoiced or have it voiced again. The dub is rebuilt with the changes.

  • Moments the software could not decide

    Each one plays in its place, original then dub, and a person accepts it with a reason or fixes the line.

  • Every version is kept

    Each published version lists the lines it changed and who changed them. An earlier text can be restored.

  • Nothing to install

    The editor runs in a browser behind a sign-in. Saved changes are stored on our side, so a closed tab loses nothing.

What stays the same

A dub should change the language and leave the conversation alone.

  • Who is talking

    Every person in the video keeps their own voice, so a conversation between three people still sounds like three people.

  • When they talk

    Dubbed lines start where the original lines start. A line that runs long is fitted to its slot, and people who talk over each other still do.

  • How loud they are

    Each dubbed line is set to the level of the line it replaces, so the balance between speakers is the one your editor mixed.

  • Laughter and music

    With separate tracks, laughs, reactions and the music bed are taken from the original recording and mixed back in.

Languages

  • Russian and English
  • German and English

Both pairs work in either direction. Each language has its own translators. If you need a pair that isn't listed, ask, and we'll tell you whether we can do it well.

What to send

Either works. Separate tracks keep more of the original.

A link to the videoSeparate tracks from your editor
Telling speakers apartWorked out from the audioKnown from each speaker's track
People talking at onceRecovered where it can be heardKept cleanly
Laughs and reactionsKept between linesKept, including over speech
Music and effectsNot keptMixed back in

Separate tracks means one audio file per speaker plus one for music and effects, lined up with the video and balanced as in your final mix.

Who it's for

Long conversations where the way people sound is part of the content: interviews, podcasts, lectures, courses and talks. A two-hour interview has been dubbed from start to finish.

We clone a voice only for a video whose owner has asked us to dub it.

Samevoice is built by Anna Shalaginov, a machine-learning engineer in Boston, Massachusetts.

Request a dub

Tell us what the video is, how long it runs and which language you want it in. We'll reply with what we need from you and what it will cost.

[email protected]