Audio Transcription
Powered by Whisper AI with Speaker Detection
Manage Custom Voices
Record a Voice
Records your computer's sound (a video, a game, a call) to use as a voice. Press record, then in the popup pick Entire screen and turn on "Also share system audio" (or pick a Chrome tab and turn on "Also share tab audio"). Only the sound is kept. Aim for 15–30 seconds of one person talking with no music. Works in Chrome or Edge.
Record your own voice. Use a quiet room, keep the mic a hand's width away, and speak naturally for 15–30 seconds. Your browser will ask for mic permission the first time.
Upload Voice Sample (for cloning)
Upload a clear WAV/MP3 of the target voice. Uses GPU server when available, CPU fallback (slower).
Upload Piper Model (instant playback)
Upload a trained .onnx + .onnx.json pair for instant local TTS.
Prepare Training Dataset
Upload raw audio → auto-transcribe with Whisper → download training-ready dataset ZIP.
Train on a GPU machine, then upload the .onnx model for instant playback (~0.5s).
Click or drag files here. Your originals are never changed; every edit makes a new file.
Separate audio & video
Pull the sound out as its own file, or get the video with no sound.
Trim, cut & split
Uses the selection and split points above.
Audio effects
Volume, loudness, fades and speed.
Combine
Join clips end-to-end, or put different audio on this file.
Convert
Save a copy in another format.
Users
| Username | Role | Status | Transcriptions | Joined | Actions |
|---|