Isolate the speaking voice in your video

Bring the speaker forward in a noisy recording. Hear an actual voice extracted from a rally video, then try your own audio or video with a free 30-second preview.

Upload a file

Free 30-second preview · No signup

Hear the difference

Hear the speech, with less of the surrounding crowd.

Man speaking plays the extracted voice from this rally recording. Switch to Original to hear the full scene at the same point in the video.

0:00
01

Upload your recording

Choose an audio or video file. Select a section where the speaker and the surrounding sounds can both be heard.

02

Describe the voice

Enter a short description of the voice in the sound field. This rally example uses "man speaking".

03

Keep the speech result

Listen to A and B separately in the preview. Leave only the result with the voice selected, then choose Process full file. You can check both results again before downloading.

Separate speech from the rest of the scene

A recording of a talk can include the speaker, nearby conversations, and crowd reactions in the same soundtrack. Voice isolation separates the described voice so you can choose the speech result. The rally above shows what that sounds like on an actual recording.

Works well on

  • Extracting the speaking voice demonstrated in the rally video above
  • Comparing a spoken passage with the surrounding sounds reduced
  • Working directly with a video, without exporting its soundtrack first
  • Checking a short sample before processing the full talk or recording

Know before you start

  • Overlapping speakers can be difficult to separate. A description does not guarantee that only one person will remain.
  • Listen for lost words, changes to the voice, and crowd sounds that remain in the extracted result.
  • Clipping, distortion, or speech missing from the recording cannot be repaired simply by separating the voice.
  • A speaking voice and a sung vocal are different targets. Use a description that matches your recording.

Prompts that work well:

FAQ

Frequently asked questions

Can I isolate a voice from a video with background noise?

The rally demo above shows an extracted speaking voice from a real video. Try a passage from your own recording and check the words you want to keep. How clearly the voice separates depends on the sounds that overlap it.

Does extracting the audio from a video also isolate the voice?

Saving a video soundtrack as an audio file keeps all its sounds together. Voice isolation separates the described speaking voice from the rest of that soundtrack.

Which result should I choose to keep the speech?

In the demo, choose Man speaking. In your own preview, compare A and B one at a time and leave the speech result selected when you process the full file. Both separated results are available to check before downloading.

Can I separate two people talking at once?

You can try a description of the voice, but overlapping speech may remain together or lose detail. The free preview lets you judge whether the result is useful for your recording before processing the full file.

Is the preview free?

The 30-second preview is free, with up to 20 previews every 24 hours and no signup. Processing the full file uses paid minutes, starting at $4.99 for 10 minutes.

Can I use an audio file or a video from my phone?

You can upload MP3, WAV, M4A, MP4, or MOV files up to 10 minutes and 1 GB. Results download as WAV or MP3, or as MP4 for video uploads.

More from EraseSound