Audio model
Voice
Cleans speech to 48 kHz
Removes background noise from speech and extends it to 48 kHz. Built for voices, not music.
This model adds nothing
Output is built from detail already present in your file. Nothing is generated, so the result stays true to what you uploaded.
Media
Audio
Maximum scale
Same size
Scale steps
Does not enlarge
What Voice is for
Voice cleans speech. Noise drops, and the result is delivered at 48 kHz so a narrow recording has room to breathe.
It is for spoken audio. Music restoration is not in this catalogue.
When to pick Voice
Interviews
A voice recorded in a room with hiss or rumble.
Phone audio
Speech that was captured narrow and noisy.
Not music
Songs and scores wait until the same restoration can be backed by two providers.
Other audio models
Frequently asked questions
What sample rate comes back?
48 kHz.
Does Voice upscale a song?
No. It is speech denoise and bandwidth extension, not music restoration.
Which files can I upload?
WAV, MP3, M4A, and FLAC, up to 25 MB.