whisper-windows-mcp

FFmpeg is required for video files and non-native audio formats.

Or download fromffmpeg.organd add to your PATH.

{ "mcpServers": { "whisper": { "command": "npx", "args": ["-y", "whisper-windows-mcp"], "env": { "WHISPER_CLI_PATH": "C:\\whisper\\Release\\whisper-cli.exe", "WHISPER_MODEL": "C:\\whisper\\models\\ggml-medium.en.bin" } } } }

Config file location:C:\Users\YourName\AppData\Roaming\Claude\claude_desktop_config.json

Save andfully restartClaude Desktop. You should seewhisperlisted with a green running badge in Settings → Developer.

This confirms your GPU is detected and Vulkan acceleration is active.

Transcribe a single file. Supports blocking (default) or background mode for long files.

- timestamps— timestamped segments, e.g.[00:00:01.230 --> 00:00:04.560] Hello world(default)
- text— plain text, no time codes
- json— structured JSON (blocking mode only)
- srt— SubRip subtitle file saved next to source
- vtt— WebVTT subtitle file saved next to source
- lrc— LRC lyrics/karaoke format saved next to source
- csv— CSV with timestamps saved next to source

Monitor a background transcription job started withtranscribe_audio(background=true).

Returns elapsed time, last processed timestamp, and the full transcript when complete.

Automated sequential batch transcription of all untranscribed files in a folder. Sorts by duration (shortest first), processes one at a time as background jobs, validates each output. Batch self-advances when each file finishes — no polling required.

Monitor a running batch. Automatically advances to the next file when the current one finishes. Returns overall progress, current file with timestamp, and any failed files.

Process files one at a time with a preview and confirmation before each. Useful when you want to review as you go.

Generate subtitle files. Supports automatic language detection and English translation output. Outputs SRT (widest compatibility) or WebVTT (web and HTML5 video).

When both native and translation are requested, two files are saved next to the source:

- filename.ja.srt— original language
- filename.en.srt— English translation

Whisper's built-in translation only translatesto English. For other target languages, translate the subtitle file contents separately.

Analyze files before committing to transcription. Returns duration, size, codec, and estimated transcription time on CPU and GPU. For folders, shows all files in a sortable table with transcription status.

Verify whisper-cli.exe, the model file, and FFmpeg are all accessible. Run this first if anything is failing.

List all Whisper model files installed in your models directory. Shows filename, size, whether it is currently active, quantization status, and recommended use case. No network calls — reads local filesystem only.

Download a Whisper model directly from Hugging Face into your models directory. Only downloads from trusted Hugging Face namespaces. After downloading, useswitch_modelto activate it.

Switch the active Whisper model for the current session without restarting Claude Desktop. Change is session-scoped — does not persist after restart. To make permanent, updateWHISPER_MODELin your config.

Detect GPU hardware and verify Vulkan acceleration is available. Reports GPU name, VRAM, whetherggml-vulkan.dllis present, and recommends the best model size for your hardware.

Start, stop, or check thepersistent model server(whisper.cpp'swhisper-server). While running, the active model stays resident in VRAM and everytranscribe_audio/transcribe_batchcall is served over localhost withno per-file model reload— a large speedup when transcribing many short files, where the one-time model-load cost otherwise dominates.

- ⚠️The resident model holds GPU VRAM for the server's whole lifetime.Start it deliberately, do your work, thenstopit to hand the GPU back to other applications sharing the card. Stopping performs a full kill so VRAM is actually released.
- switch_modelwhile the server is running hot-swaps the resident model in place (no restart).
- Bound to127.0.0.1only — never exposed on the network.
- While the server is up, operations that need the one-shot CLI — background jobs,start_batch,generate_subtitles,lrc/csvoutput, and advanced per-call options the HTTP API doesn't honor (beam_size,best_of,word_timestamps,diarize,tinydiarize,vad_model,offset_t,duration) — arerefusedwith a "stop the server first" message rather than silently ignored, so no second engine ever contends for the GPU.
- Requireswhisper-server.exe(ships alongsidewhisper-cli.exe). Configure withWHISPER_SERVER_PATH/WHISPER_SERVER_PORTif needed.

The pre-built Vulkan release enables GPU acceleration automatically. Tested on AMD Radeon RX Vega 56 (GCN 5th gen). Any GPU with Vulkan 1.0+ support should work, including NVIDIA and Intel Arc.

Performance comparison (large-v3 model, ~14 minute audio file):

GPU utilization during transcription is typically 15–20%, dropping back to idle between files.

Supports Windows 10 and Windows 11. No Windows 11-specific configuration is required — the tool makes no Win32 API calls and runs on either OS.

Whisper can auto-detect the spoken language and transcribe in that language. The built-in translation model translatesto English only.