0
Speakers max, auto-detected from a single track.
The native AI plugin for DaVinci Resolve·2022 +Blackmagic official partner1 audio.Every voice, isolated.
Diarization automatically detects each speaker in an audio track and splits them into dedicated DaVinci Resolve lanes (A1, A2, A3...). Native split, up to 10 voices, free.
Drop a single audio track in. DavinciClaude detects every voice and splits it into one lane per speaker, ready for Podcast Multicam, individual EQ, or per-speaker ducking. Free.
How it works
Three steps. Split, ready to mix.
- ● 01
Drop
Drop a single audio track.
Pick any audio clip on your timeline, or import a fresh WAV/MP3. Diarization handles up to 90 minutes per pass.
A1 · source● drop heresource.wav - ● 02
Detect
AI detects every voice.
Up to 10 speakers identified automatically. Review the detected voices in the editor: rename, merge, or split a speaker before applying. Crosstalk handled gracefully.
Diarization · analyzing● 3 voicesSpeaker 1Speaker 2Speaker 3 - ● 03
Split
One lane per speaker.
Each speaker lands on their own audio track (A1, A2, A3...) on your real DaVinci Resolve timeline. Now you can run Podcast Multicam autocut, EQ each voice individually, or duck the music under each speaker.
Premiere timeline● split doneA1A1S1A2S2A3S3
What ships
The prep step before Podcast Multicam.
Free
Free with daily quota. Pro+ at $7.99/mo.
0‰
Turn-detection accuracy on clean recordings.
A1, A2, A3, ...
Each detected voice lands on its own native DaVinci Resolve audio track. Ready for Podcast Multicam, individual EQ, or per-speaker ducking.
CH · DaVinci Resolve
Native audio tracks. Edit like any source.
Stack the rest
Explore other tools.
- Claude · Ask anything
Plain-English prompts that edit your real timeline.
Learn more - Vibe Motion
Motion design generated from a single prompt.
Learn more - GenAI
Any GenAI model, native in DaVinci Resolve. By Fal.AI.
Learn more - Smart Captions
Word-by-word, animated captions in 99 languages.
Learn more - Podcast · Multicam
Active speaker detection, multicam autocut.
Learn more - Smart Silences
Auto-detect silences and tighten the edit.
Learn more - Claude Cut
Removes bad takes & repetitions, follows your script.
Learn more - Smart Virals
Turn 1h of content into 10 vertical viral clips.
Learn more - Smart Subtitles
0 spelling mistakes. Claude corrected, native subtitles.
Learn more - Auto Chapters
Animated MoGRT chapters from a single prompt.
Learn more - Auto Zoom
Emotion-driven V2 zoom clips, generated.
Learn more
From the Discord
Real users, real reactions.
FAQ
What you're probably wondering.
It splits a single audio track into one lane per detected voice. Most useful before running Podcast Multicam, since multicam autocut needs each speaker on their own track. Also handy for individual EQ, noise reduction, or ducking per speaker.
Up to 10. Auto-detection picks up overlapping speech and short interventions; you can manually merge or rename speakers in the editor before applying.
Built on a state-of-the-art diarization model with 99.5% turn-detection accuracy on clean recordings. Quality drops on heavy crosstalk, but you can fine-tune any lane afterwards on the DaVinci Resolve timeline.
Yes, voice separation is language-agnostic. 99 languages supported for the optional transcript output.
Diarization runs on our servers (encrypted in transit, deleted right after). No silent uploads, no model training on your content.
DaVinci Resolve 2022 and later, on macOS (Intel + Apple Silicon) and Windows.
REC
SP · HiFi · CH 03
▶ PLAY
00:14:32:08