Transcribe File

Transcribe File

This mode is for offline or recorded media, not live dictation.

Add a File

You can:

• drag and drop a supported audio/video file
• click Choose File

Supported extensions are shown in the info tooltip on the drop zone.

Transcription Controls

For selected file:

• choose transcription model
• choose language (when model supports manual language choice)

Speaker Diarization (Local Models)

When local model provider is active, you can enable:

• Speaker Diarization
• Enhanced Diarization

You can also set default speaker labels before processing.

Use diarization when you need "who said what" structure.

Spark After Transcription

TranscribeFileSparkEnabled controls whether Spark runs after transcription.

When enabled, you can choose:

• Spark prompt
• Spark provider
• Spark model

These are stored as Transcribe File-specific preferences and can differ from your global Spark defaults.

Output Review

After processing:

• review Sparked text (if Spark enabled)
• review Original text
• copy or save either output
• adjust transcript font size in viewer

Recommended Workflow

1. transcribe without Spark first
2. check recognition quality and speaker segmentation
3. enable Spark for formatting/summarization once raw transcript quality is acceptable

Common Issues

Speaker labels are messy

Retest with local diarization and cleaner source audio.

Spark output over-edits details

Use a stricter prompt that preserves names, numbers, and quoted statements.