Bulk Transcribe

Local-first transcription with a review workflow built for 100% accuracy

Server Mode (session-isolated)

Import

Upload Audio Files

Drop audio files here or click to upload

Supports: MP3, WAV, M4A, FLAC, OGG, WebM, MP4, MOV, AVI

Model

Language

Output Options Saved

Segmented transcription Full transcription

Include timestamps Word-level timestamps Overwrite existing files

Processing mode

Parallel workers

Metal GPU: ~5-6x realtime on Apple Silicon. Single worker only.

Speaker Diarization

Identifies who spoke when. Requires HuggingFace token (HF_TOKEN).

Min speakers

Max speakers

Fixed count

Tip: Set "Fixed count" if you know the exact number of speakers.

Max duration per segment

Auto-split long audio

Splits files longer than max duration into overlapping chunks.

For ~25 minute files, 15 min with auto-split works well.

Fast Switching (short segments, higher overlap)

Quality preset

Silence threshold

Max segment length

Max segment length only applies when word timestamps are enabled.

Audio Pre-processing

VAD pre-filter Filters silence before Whisper

Hallucination detection Flags suspicious segments

Sensitivity 50%

Keep more Filter more

Active Job

RUNNING

Started

--:--:--

Elapsed

00:00

File 0 of 0 0%

Starting...

Activity

Review Transcript

Saved

Done

Batch

File Loading...

0:00 / 0:00

Speakers

Keyboard Shortcuts

Space Play/pause

[ ] Prev/next chunk

1-9 Assign speaker

E Edit chunk

H Highlight chunk

⌘/Ctrl+S Split at cursor

⌘/Ctrl+⇧+S Split interjection

Enter Save edit

Esc Cancel edit

Backspace Delete chunk

⌘/Ctrl+Z Undo

0 chunks

No chunk selected

Merged Auto-scroll

Loading timeline...

Bulk Transcribe

Model Manager

Active Job

Results

Edit Speakers

Speakers

Merge Speakers

Import Project

Add Missing Chunk