AI Speaker Separation Split Voices from One Recording

Yes. Seply splits spoken conversations into a separate, time-aligned WAV track for each detected speaker. Upload your podcast, interview, or recording, then preview each voice and download the tracks you need. It is designed for speech, not separating singers from a song.

Advanced Overlap is designed for interruptions and simultaneous speech and requires a successful purchase. Use Standard for ordinary turn-taking conversations. Results vary with noise, echo, similar voices, and the amount of overlap; some of the other voice may remain.

Yes. First separate the conversation, then listen to the resulting tracks to find the person you want and download only their WAV. Seply does not ask you to identify a person by name or provide a voice sample before separation. The track keeps the original timing, so pauses can remain.

You get separate WAV audio tracks for the detected speakers. Import the voices you need into your video editor and mute the original mixed audio to avoid playing it twice. This workflow does not export an edited or dubbed video.

New accounts receive 30 one-time trial credits, valid for 30 days. They cover up to about 3 minutes of Standard Separation, at 10 credits per started minute with a 10-credit minimum. Advanced Overlap requires a successful purchase and costs 60 credits per started minute with a 60-credit minimum. The exact estimate appears before processing.

No. Speaker diarization labels who spoke when; transcription adds the words. Speaker separation creates an individual audio track for each detected speaker. Use Seply's separate Speech to Text tool when you need a transcript with speaker labels.

Standard supports the formats listed in the upload area, including WAV, MP3, M4A, AAC, FLAC, OGG, OPUS, WMA, Speex, MP4, AVI, MOV, MKV, and WebM. Advanced Overlap accepts WAV, FLAC, MP3, AAC, M4A, MP4, and MOV files up to 90 minutes; MP4 and MOV files must contain AAC or PCM audio.

Yes. Seply creates an independent, time-aligned WAV file for each detected speaker.

The server-calculated media duration is rounded to the nearest whole second before billing. Standard Separation uses 10 credits per started minute with a 10-credit minimum, so a displayed 1:00 recording uses 10 credits and 1:01 uses 20. Advanced Overlap uses 60 credits per started minute with a 60-credit minimum, so displayed 1:00 and 1:01 recordings use 60 and 120 credits respectively.

Monthly plan credits expire 30 days after they are granted. One-time credit packs expire 12 calendar months after purchase. Canceling a subscription does not remove credits before their own expiration dates.

Credits reserved for a confirmed failed job are automatically returned to the original credit batches with their original expiration dates.

For new jobs, the original recording and completed speaker tracks are stored privately for 30 days from completion. You can listen in your history and download separated tracks during that period. Limited job metadata remains after media expires.

You may request a refund within 7 calendar days for purchased credits that have not been used. Used credits and completed jobs are generally not refundable. Exceptions apply for duplicate charges, failed delivery, and rights required by law.