The speech model is not ready
Open Settings and let Desktop finish downloading and loading the local model before retrying.
LOCAL SPEECH TO TEXT
Local Video Transcription on Windows and macOS
Create TXT transcripts and SRT subtitles from downloaded video on your own computer. Desktop prepares audio with FFmpeg and runs the speech model without uploading the media.
- Local speech recognition
- TXT + timestamped SRT
- No media upload
Support & limits
What offline transcription handles
Desktop prepares 16 kHz audio with FFmpeg, detects speech, then recognizes one segment at a time with the local model.
- Supported input
- A downloaded video task with Extract Text & Subtitles selected. Supported speech includes Mandarin, Cantonese, English, Japanese and Korean.
- Free
- Available for compatible downloaded video tasks; Free has 10 successful single-video parses per day and no profile extraction.
- Desktop Pro
- Unlimited single-video parsing and profile extraction; transcription still runs on the device.
- Profile paging
- Transcription has no paging. In a Pro profile batch, only accessible queued video tasks can be transcribed.
- Outputs
- Both UTF-8 TXT and timestamped SRT are written beside the compatible downloaded media.
- Systems
- Windows 64-bit; macOS Apple Silicon and Intel. Speed depends on the local CPU and media duration.
Actual workflow
How to create local transcripts
Each step maps to a control or task state in the Desktop application shown above.
- 01
Select Extract Text & Subtitles
Enable TXT + SRT before starting a compatible video download.
- 02
Enable the local model
On first use, confirm setup and wait until Settings reports Enabled or Ready.
- 03
Let the task finish
Desktop downloads the video, prepares audio and processes transcription serially.
- 04
Open the outputs
Find the matching .txt and .srt files beside the downloaded media.
When a task stops
Common failure reasons
Failures remain attached to their task so completed items in the same queue stay available.
No speech is detected
Confirm the video contains audible speech. Music-only, silent or extremely noisy audio can produce empty files.
Audio preparation fails
Keep the downloaded video available and retry. A damaged or unsupported file may prevent FFmpeg from preparing audio.
Windows · macOS
Get SnapVideoTools Desktop
Choose the current package for your computer. Release cards appear only when that package is available.
WINDOWSWindows 64-bitx64 · .EXE · v1.0.2↓MACOSApple SiliconARM64 · .DMG · v1.0.2↓MACOSIntel Macx64 · .DMG · v1.0.2↓
Free: 10 successful single-video parses per day; no profile extraction. Desktop Pro: $9.90/month or $79.90/year for unlimited parsing and supported profile extraction. Compare plans.
FAQ
Questions about this workflow
Is the video uploaded for transcription?
No. Audio preparation and speech recognition run locally on the computer.
Which languages are supported?
The current Settings screen lists Mandarin, Cantonese, English, Japanese and Korean.
Is accuracy guaranteed?
No. Results depend on language, speech clarity, background noise, overlap and the source recording.