How omar-abuhadhoud/transcriber works
Transcriber is a local-first Windows desktop app for downloading media audio and transcribing it on the user’s GPU. The main workflow is to add local files—or download audio from a supported link into the queue—then run queued items one at a time and use the resulting text. Recovery files preserve partial output; completed transcripts can be copied, saved, or opened with Arabic RTL formatting. The README also documents installation and update capabilities. Some runtime connections are supported only by README or source-index references, so uncertain links are omitted or marked dashed.
Main entry points: queue_page.py Transcription queue · download_page.py Download page · app.py App shell · media_item.py Media row
Components 25
Desktop experience
- App shell
ctk_ui/app.py - Transcription queue
ctk_ui/queue_page.py - Media row
ctk_ui/media_item.py - Transcript actions
ctk_ui/media_item.py
Media downloads
- Download page
ctk_ui/download_page.py - Link dialog
ctk_ui/url_dialog.py - Download pool
downloader/pool.py - Provider registry
downloader/registry.py - Downloader contract
downloader/base.py - Audio fetch
downloader/ytdlp.py - Safe naming
downloader/naming.py
Transcription runtime
- Transcription worker
transcriber/transcribe_module.py - Engine registry
transcriber/registry.py - Qwen ASR
transcriber/engines/qwen_asr.py - Audio handling
transcriber/audio.py - Speech segmentation
transcriber/vad.py - Chunk logging
transcriber/chunk_log.py
Local state
- Recovery files
global_vars.py - Local paths
transcriber/paths.py
Install and updates
- Runtime provisioning
installer/provision.py - Install state
transcriber/install_state.py - Update service
transcriber/update.py
Other components
- User
- Model downloads
- Release service
Connections 29
- User → App shell: uses
- App shell → Transcription queue: builds
ctk_ui/app.py - App shell → Download page: builds
ctk_ui/app.py - User → Transcription queue: adds media
- Transcription queue → Media row: creates rows
ctk_ui/queue_page.py - Media row → Transcription queue: requests work
ctk_ui/media_item.py - Transcription queue → Transcription worker: dispatches jobs
ctk_ui/queue_page.py - Transcription worker → Engine registry: selects engine
- Engine registry → Qwen ASR: provides engine
- Qwen ASR → Audio handling: processes audio
transcriber/engines/qwen_asr.py - Qwen ASR → Speech segmentation: segments speech
transcriber/engines/qwen_asr.py - Qwen ASR → Chunk logging: records chunks
transcriber/engines/qwen_asr.py - Media row → Recovery files: streams text
ctk_ui/media_item.py - Media row → Transcript actions: offers actions
ctk_ui/media_item.py - User → Download page: pastes link
- Download page → Link dialog: opens
ctk_ui/download_page.py - Link dialog → Download pool: submits job
ctk_ui/download_page.py - Download pool → Provider registry: resolves provider
downloader/pool.py - Download pool → Downloader contract: runs downloader
downloader/pool.py - Downloader contract → Audio fetch: delegates fetching
downloader/providers/youtube.py - Audio fetch → Safe naming: names output
downloader/ytdlp.py - Download page → Transcription queue: adds finished media
ctk_ui/app.py - Download page → Local paths: uses download paths
ctk_ui/download_page.py - Recovery files → Local paths: locates data
global_vars.py - Runtime provisioning → Install state: records provisioning
installer/provision.py - Runtime provisioning → Model downloads: fetches weights
- App shell → Update service: launches update
ctk_ui/app.py - Update service → Release service: checks releases
- Update service → Install state: checks installation
transcriber/update.py