Not in the top 1,000 by GitHub stars

Mobiss11/Whisper-Skill

🎤 Локальная транскрибация через Whisper без OpenAI API. Push-to-talk диктовка, вшитые сабы в MP4 (CapCut-стиль), мастер установки одной командой. Mac/Linux/Windows, 99 языков, бесплатно.

Engineering

History

Review weekly changes to SKILL.md before updating. Dates mark the weeks in which we observed each version; checks apply to the recorded commits.

View weekly changes

`SKILL.md`

Week of Nov 2, 2026 · +0 / −0 lines

Before Network or broad shell → After Network or broad shell

7a705b58d6bf → 7a705b58d6bf

Compare on GitHub

Week of Oct 26, 2026 · Line counts unavailable

Before Not checked yet → After Network or broad shell

First observation → 7a705b58d6bf

No earlier version was observed for comparison.

Compare on GitHub

View on GitHub Add to my skills

Verified commit 7a705b58d6bf4fede1f4af8e392fac09acd4b2cb

18GitHub stars

1star in 7 days

1skill in this repository

0commits in 30 days

Heat

1.8

Star growth, last 7 daysNot available
Star growth, last 24 hoursNot available
Rank movement, last 7 daysNot available
Commits, last 30 daysNot available
Accounts asking about it through MCP, last 7 daysNot available
Hot votes this weekNot available

Heat never changes the order by stars, and sponsorship never touches it. How heat works

Stars over time

30-day change not yet available: not enough history.

Individual skills in this repository

whisper-skill /

How it works

whisper-skill

When your agent uses it

Используй этот скилл когда пользователю нужна локальная транскрибация аудио/видео через Whisper, голосовой ввод (диктовка вместо клавиатуры, push-to-talk hotkey), или вшивание субтитров в MP4 в стиле CapCut. Активируй при задачах "транскрибируй файл/папку/TikTok/YouTube/подкаст", "сабы для shorts/reels/тиктока", "speech-to-text локально", "speaker diarization", "диктовка голосом", "Superwhisper alternative", "Wispr Flow alternative", "voice dictation". Скилл автоматически определяет ОС/CPU/GPU и подбирает оптимальный бэкенд (faster-whisper / whisper.cpp / whisperx / mlx-whisper / openvino) и модель под железо пользователя. Для Intel Core Ultra / Intel Arc — openvino задействует iGPU+NPU. Включает interactive wizard (wizard.py) который ставит всё одной командой. Все языки, бесплатно после установки, никаких API.

What it loads into context

Frontmatter

239 tokens

Always available to the agent

SKILL.md

3,778 tokens

When the skill is activated

Reference documents

38,697 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

Как пользоваться
Общие правила
Под-документы
Типичные ошибки и как их чинить
После транскрибации — что дальше?
Что вне скоупа

Scripts executed separately

examples/__init__.py
examples/bake_subs.py
examples/batch_folder.py
examples/common.py
examples/from_url.py
examples/podcast_diarize.py
examples/transcribe_one.py
examples/voice_dictation.py
launcher/voice_dictation_silent.vbs
scripts/build_icon.py
scripts/convert_openvino.py
scripts/cursor_indicator.py
scripts/detect_env.py
tools/install_autostart.ps1
tools/install_shortcut.ps1
tools/uninstall_autostart.ps1
wizard.py

Files

.gitignoreasset451 bytes
LICENSEasset1,065 bytes
README.mddoc28,237 bytes
SKILL.mdcore17,231 bytes
SKILL.mdcore17,231 bytes
SKILL.mdcore17,231 bytes
SKILL.mdcore17,231 bytes
SKILL.mdcore17,231 bytes
assets/icon.icoasset44,199 bytes
assets/icon.pngasset133,494 bytes
backends/faster-whisper.mddoc8,367 bytes
backends/mlx-whisper.mddoc5,839 bytes
backends/openvino.mddoc9,802 bytes
backends/whisper-cpp.mddoc7,363 bytes
backends/whisperx.mddoc6,763 bytes
docs/cost-comparison.mddoc6,103 bytes
docs/diarization.mddoc8,151 bytes
docs/installation-linux.mddoc5,913 bytes
docs/installation-mac.mddoc4,928 bytes
docs/installation-windows.mddoc7,258 bytes
docs/known-issues.mddoc11,018 bytes
docs/speed-tuning.mddoc8,714 bytes
docs/subtitle-baker.mddoc9,267 bytes
docs/voice-dictation.mddoc18,333 bytes
examples/__init__.pyruns0 bytes
examples/bake_subs.pyruns12,565 bytes
examples/batch_folder.pyruns3,914 bytes
examples/common.pyruns20,142 bytes
examples/from_url.pyruns2,702 bytes
examples/podcast_diarize.pyruns5,055 bytes
examples/transcribe_one.pyruns3,198 bytes
examples/voice_dictation.pyruns55,399 bytes
launcher/voice_dictation_silent.vbsruns1,705 bytes
methodology/quality-vs-speed.mddoc8,521 bytes
models/README.mddoc5,957 bytes
scripts/build_icon.pyruns957 bytes
scripts/convert_openvino.pyruns1,600 bytes
scripts/cursor_indicator.pyruns12,445 bytes
scripts/detect_env.pyruns32,223 bytes
tools/install_autostart.ps1runs1,314 bytes
tools/install_shortcut.ps1runs1,938 bytes
tools/uninstall_autostart.ps1runs538 bytes
wizard.pyruns20,287 bytes

Read SKILL.md


# Whisper Skill — локальная транскрибация без OpenAI API

Этот скилл учит Claude правильно ставить и использовать Whisper локально — на ноуте, маке, серваке, без всякого OpenAI API. Поддерживает все ~99 языков (Whisper мультиязычен из коробки), все ОС, и автоматически подбирает оптимальный путь под конкретное железо.

> 💰 **Сколько экономим**: OpenAI Whisper API стоит **$0.006/мин**. Если транскрибируешь 5 часов в день — это **$108/мес**. Локальная установка — **$0** после первой загрузки модели (~3 GB).

## Как пользоваться

### Шаг 0 — Самый простой путь: интерактивный мастер

Если пользователь не хочет читать документацию — **запусти мастер**, он спросит что хочешь делать и сам всё поставит:

```bash
python wizard.py

View the complete source on GitHub

Skills included

whisper-skill

Repository facts

License: MIT
Created: Apr 30, 2026
Last pushed: Aug 8, 2026
Forks: 15
Open issues: 0
Project activity: Slowing

Before you install

Network or broad shell

Automated signals, not a guarantee

Checked Nov 4, 2026

Not fully analyzed: some files were too long, unreadable or missing. This never lowers the level.

Show 4 notes

SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed

No broad tool permissions
29 dangerous shell commandsin 13 files: docs/installation-linux.md:9 scripts/detect_env.py:517 backends/faster-whisper.md:39 and 10 more
167 network referencesin 21 files: wizard.py:233 docs/installation-windows.md:26 docs/installation-linux.md:36 and 18 more
No hidden or obfuscated text
No instruction overrides
17 executable scriptsin 17 files: examples/__init__.py:1 examples/bake_subs.py:1 examples/batch_folder.py:1 and 14 more
37 references to secretsin 13 files: examples/common.py:51 examples/voice_dictation.py:75 examples/podcast_diarize.py:10 and 10 more
No requests to vote in Hot

Show all 250 signals by file

`README.md`

Network accessline 16

git clone https://github.com/Mobiss11/Whisper-Skill.git ~/.claude/skills/whisper-skill

Network accessline 49

1. Скачай ZIP-архив: https://github.com/Mobiss11/Whisper-Skill/archive/refs/heads/main.zip

Network accessline 69

# 1. Склонируй git clone https://github.com/Mobiss11/Whisper-Skill.git

Network accessline 70

git clone https://github.com/Mobiss11/Whisper-Skill.git

Network accessline 234

# TikTok python -m examples.from_url "https://www.tiktok.com/@user/video/1234567890123456789"

Network accessline 235

python -m examples.from_url "https://www.tiktok.com/@user/video/1234567890123456789"

Network accessline 237

# YouTube с указанием языка python -m examples.from_url "https://www.youtube.com/watch?v=dQw4w9WgXcQ" --language en

Network accessline 238

python -m examples.from_url "https://www.youtube.com/watch?v=dQw4w9WgXcQ" --language en

Network accessline 240

# Instagram Reel в конкретную папку, без удаления mp3 python -m examples.from_url "https://www.instagram.com/reel/Cv1AbcXXXX/" --output ./content/ --keep-audio

Network accessline 241

python -m examples.from_url "https://www.instagram.com/reel/Cv1AbcXXXX/" --output ./content/ --keep-audio

Network accessline 247

# Получить HF token (бесплатно): https://huggingface.co/settings/tokens

Network accessline 248

# Принять условия: https://huggingface.co/pyannote/speaker-diarization-3.1

Access to secretsline 249
```
[Credential-related source omitted]
```

Dangerous shell operationline 524

Шрифт не установлен в системе. На Mac: `brew install --cask font-montserrat font-inter font-impact font-roboto`. На Linux: `sudo apt install fonts-montserrat fonts-inter fonts-roboto`. Или поменяй в `examples/bake_subs.py` стиль на тот шриф

`SKILL.md`

Network accessline 80

# Из URL (TikTok/YouTube/Reels) → транскрибат через yt-dlp python -m examples.from_url "https://www.tiktok.com/@user/video/123..."

Network accessline 81

python -m examples.from_url "https://www.tiktok.com/@user/video/123..."

`backends/faster-whisper.md`

Network accessline 3

**GitHub:** https://github.com/SYSTRAN/faster-whisper

Network accessline 4

**PyPI:** https://pypi.org/project/faster-whisper/

Network accessline 29

# Если nvcc не нашёл — поставь CUDA Toolkit: # https://developer.nvidia.com/cuda-downloads (выбери ОС → runfile или deb)

Network accessline 30

# https://developer.nvidia.com/cuda-downloads (выбери ОС → runfile или deb)

Dangerous shell operationline 39
```
sudo apt install -y ffmpeg
```

Network accessline 45

# CUDA Toolkit + driver: https://developer.nvidia.com/cuda-downloads

Network accessline 76

# Сначала ROCm: https://rocm.docs.amd.com/projects/install-on-linux/ # Затем PyTorch с ROCm wheel: pip install torch --index-url https://download.pytorch.org/whl/rocm6.0

Network accessline 77

# Затем PyTorch с ROCm wheel: pip install torch --index-url https://download.pytorch.org/whl/rocm6.0 pip install faster-whisper

Network accessline 78

pip install torch --index-url https://download.pytorch.org/whl/rocm6.0

Network accessline 202

# Linux: sudo apt install libcudnn8 libcudnn8-dev # Или скачай вручную: https://developer.nvidia.com/cudnn

Dangerous shell operationline 203

sudo apt install libcudnn8 libcudnn8-dev

Network accessline 204

# Или скачай вручную: https://developer.nvidia.com/cudnn

Network accessline 209

Не хватает CUDA Toolkit. Установи: https://developer.nvidia.com/cuda-downloads

`backends/mlx-whisper.md`

Network accessline 3

**GitHub:** https://github.com/ml-explore/mlx-examples

Network accessline 4

**PyPI:** https://pypi.org/project/mlx-whisper/

`backends/openvino.md`

Network accessline 3

**GitHub:** https://github.com/openvinotoolkit/openvino

Network accessline 4

**PyPI:** https://pypi.org/project/optimum-intel/

Network accessline 44

Открой Device Manager → Display adapters → правый клик на Intel(R) Arc → Properties → Driver tab. Версия должна быть **32.0.x** или новее (релизы 2024+). Старые драйверы не поддерживают Whisper-операции на iGPU. Обновить: https://www.intel.

Dangerous shell operationline 50

sudo apt install -y intel-opencl-icd intel-level-zero-gpu level-zero

Network accessline 50

sudo apt install -y intel-opencl-icd intel-level-zero-gpu level-zero # NPU драйвер: # https://github.com/intel/linux-npu-driver/releases

Network accessline 52

# https://github.com/intel/linux-npu-driver/releases

Access to secretsline 115
```
[Credential-related source omitted]
```
Access to secretsline 116
```
[Credential-related source omitted]
```

`backends/whisper-cpp.md`

Network accessline 3

**GitHub:** https://github.com/ggerganov/whisper.cpp

Network accessline 28

# Скачай модель в формате GGML mkdir -p ~/whisper-models && cd ~/whisper-models curl -LO https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-large-v3-turbo.bin

Network accessline 30

curl -LO https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-large-v3-turbo.bin

Network accessline 36

# Сборка из исходников (без GPU) sudo apt install -y build-essential cmake ffmpeg git clone https://github.com/ggerganov/whisper.cpp.git

Dangerous shell operationline 37

sudo apt install -y build-essential cmake ffmpeg

Network accessline 37

sudo apt install -y build-essential cmake ffmpeg git clone https://github.com/ggerganov/whisper.cpp.git cd whisper.cpp

Network accessline 38

git clone https://github.com/ggerganov/whisper.cpp.git

Network accessline 59

# Поставь CMake + Visual Studio Build Tools, затем: git clone https://github.com/ggerganov/whisper.cpp.git

Network accessline 60

git clone https://github.com/ggerganov/whisper.cpp.git

`backends/whisperx.md`

Network accessline 3

**GitHub:** https://github.com/m-bain/whisperX

Network accessline 4

**PyPI:** https://pypi.org/project/whisperx/

Dangerous shell operationline 35
```
sudo apt install -y ffmpeg
```

Network accessline 46

1. Зайти на HF: https://huggingface.co/pyannote/speaker-diarization-3.1

Network accessline 47

2. Согласиться с условиями использования (бесплатно) 3. То же самое для https://huggingface.co/pyannote/segmentation-3.0 4. Создать токен: https://huggingface.co/settings/tokens (тип "Read")

Network accessline 48

3. То же самое для https://huggingface.co/pyannote/segmentation-3.0

Network accessline 49

4. Создать токен: https://huggingface.co/settings/tokens (тип "Read")

Access to secretsline 50
```
[Credential-related source omitted]
```
Access to secretsline 73
```
[Credential-related source omitted]
```

Network accessline 131

Не принял условия на HF. Зайди на https://huggingface.co/pyannote/segmentation-3.0 и согласись. Потом — то же для `pyannote/speaker-diarization-3.1`.

`docs/cost-comparison.md`

Network accessline 121

- **Groq Whisper** (https://groq.com/) — Whisper-large-v3 на их LPU чипах. **Бесплатный tier** есть. Платный ~$0.00115/мин (в 5 раз дешевле OpenAI). Но всё равно платно.

`docs/diarization.md`

Access to secretsline 41
```
[Credential-related source omitted]
```

Network accessline 65

1. Зарегайся на https://huggingface.co

Network accessline 66

2. Согласись на use-conditions:    - https://huggingface.co/pyannote/speaker-diarization-3.1    - https://huggingface.co/pyannote/segmentation-3.0

Network accessline 67

- https://huggingface.co/pyannote/speaker-diarization-3.1

Network accessline 68

- https://huggingface.co/pyannote/segmentation-3.0

Network accessline 69

3. Создай токен на https://huggingface.co/settings/tokens (тип "Read")

Access to secretsline 72
```
[Credential-related source omitted]
```

Network accessline 125

# Дальше при diarization сравнивать с базой через cosine similarity # (см. https://github.com/pyannote/pyannote-audio/discussions)

Network accessline 126

# (см. https://github.com/pyannote/pyannote-audio/discussions)

Access to secretsline 157
```
[Credential-related source omitted]
```

Network accessline 185

- **simple-diarizer** (https://github.com/cvqluu/simple_diarizer) — чисто на pyannote, без whisperx, проще

`docs/installation-linux.md`

Dangerous shell operationline 9
```
sudo apt update
```

Dangerous shell operationline 10

sudo apt install -y python3 python3-venv python3-pip ffmpeg git curl

Dangerous shell operationline 15

sudo pacman -S python python-pip ffmpeg git curl

Dangerous shell operationline 20

sudo dnf install -y python3 python3-pip ffmpeg-free git curl

Dangerous shell operationline 36

sudo apt install -y nvidia-driver-535   # Ubuntu

Network accessline 36

sudo apt install -y nvidia-driver-535   # Ubuntu # или через инсталлятор: https://www.nvidia.com/Download/index.aspx

Network accessline 37

# или через инсталлятор: https://www.nvidia.com/Download/index.aspx

Network accessline 43

# Ubuntu 22.04 — пример для CUDA 12.4: wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-keyring_1.1-1_all.deb

Network accessline 44

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-keyring_1.1-1_all.deb

Dangerous shell operationline 45
```
sudo dpkg -i cuda-keyring_1.1-1_all.deb
```
Dangerous shell operationline 46
```
sudo apt update
```
Dangerous shell operationline 47
```
sudo apt install -y cuda-toolkit-12-4
```

Network accessline 50

Или скачай runfile с https://developer.nvidia.com/cuda-downloads — выбери ОС.

Dangerous shell operationline 61

sudo apt install -y libcudnn8 libcudnn8-dev

Dangerous shell operationline 108

sudo apt install -y build-essential cmake

Network accessline 108

sudo apt install -y build-essential cmake git clone https://github.com/ggerganov/whisper.cpp.git

Network accessline 109

git clone https://github.com/ggerganov/whisper.cpp.git

Network accessline 124

# Ubuntu 22.04 — см. https://rocm.docs.amd.com/projects/install-on-linux/ wget https://repo.radeon.com/amdgpu-install/6.0.2/ubuntu/jammy/amdgpu-install_6.0.60002-1_all.deb

Network accessline 125

wget https://repo.radeon.com/amdgpu-install/6.0.2/ubuntu/jammy/amdgpu-install_6.0.60002-1_all.deb

Dangerous shell operationline 126

sudo apt install ./amdgpu-install_6.0.60002-1_all.deb

Dangerous shell operationline 127
```
sudo amdgpu-install --usecase=rocm
```

Network accessline 136

python3 -m venv .venv && source .venv/bin/activate pip install torch torchaudio --index-url https://download.pytorch.org/whl/rocm6.0

Network accessline 137

pip install torch torchaudio --index-url https://download.pytorch.org/whl/rocm6.0

Dangerous shell operationline 150

sudo apt install -y libvulkan-dev vulkan-tools

Network accessline 152

git clone https://github.com/ggerganov/whisper.cpp.git

Network accessline 166

# HF token (бесплатно): # https://huggingface.co/settings/tokens

Network accessline 167

# https://huggingface.co/settings/tokens

Access to secretsline 169
```
[Credential-related source omitted]
```

Network accessline 177

# Прокси для pip pip install --proxy http://proxy:port faster-whisper

Network accessline 178

pip install --proxy http://proxy:port faster-whisper

`docs/installation-mac.md`

Network accessline 10

# 1) Homebrew если ещё нет /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

Dangerous shell operationline 11

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

Network accessline 11

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

Network accessline 45

mkdir -p ~/whisper-models cd ~/whisper-models curl -LO https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-large-v3-turbo.bin

Network accessline 47

curl -LO https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-large-v3-turbo.bin

Network accessline 90

# Hugging Face token (бесплатно) # 1. Зарегайся на https://huggingface.co

Network accessline 91

# 1. Зарегайся на https://huggingface.co

Network accessline 92

# 2. Создай токен https://huggingface.co/settings/tokens

Network accessline 93

# 3. Согласись с условиями этих двух моделей: #    https://huggingface.co/pyannote/speaker-diarization-3.1 #    https://huggingface.co/pyannote/segmentation-3.0

Network accessline 94

#    https://huggingface.co/pyannote/speaker-diarization-3.1

Network accessline 95

#    https://huggingface.co/pyannote/segmentation-3.0

Access to secretsline 97
```
[Credential-related source omitted]
```

Network accessline 116

# Скачать sample curl -LO https://github.com/ggerganov/whisper.cpp/raw/master/samples/jfk.wav

Network accessline 117

curl -LO https://github.com/ggerganov/whisper.cpp/raw/master/samples/jfk.wav

Network accessline 123

yt-dlp -x --audio-format mp3 -o sample.%\(ext\)s "https://www.youtube.com/watch?v=KAZdTHMmxYw"

`docs/installation-windows.md`

Dangerous shell operationline 17
```
sudo apt update
```

Dangerous shell operationline 18

sudo apt install -y python3 python3-venv python3-pip ffmpeg git

Network accessline 26

nvidia-smi # Если работает — всё ок # Если не работает — обнови NVIDIA driver на Windows: https://www.nvidia.com/Download/index.aspx

Network accessline 28

# Если не работает — обнови NVIDIA driver на Windows: https://www.nvidia.com/Download/index.aspx

Network accessline 40

# 3) Если есть NVIDIA GPU — CUDA Toolkit # https://developer.nvidia.com/cuda-downloads → Windows → exe (network)

Network accessline 41

# https://developer.nvidia.com/cuda-downloads → Windows → exe (network)

Network accessline 43

# 4) Build Tools (нужны для некоторых wheels) # https://visualstudio.microsoft.com/visual-cpp-build-tools/

Network accessline 44

# https://visualstudio.microsoft.com/visual-cpp-build-tools/

Network accessline 80

# Через chocolatey (поставь его если нет: https://chocolatey.org)

Network accessline 83

# Или скачай готовый бинарник: https://github.com/ggerganov/whisper.cpp/releases

Network accessline 88

# Если есть curl curl -LO https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-large-v3-turbo.bin

Network accessline 89

curl -LO https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-large-v3-turbo.bin

Network accessline 91

# Или через PowerShell Invoke-WebRequest -Uri "https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-large-v3-turbo.bin" -OutFile "ggml-large-v3-turbo.bin"

Network accessline 92

Invoke-WebRequest -Uri "https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-large-v3-turbo.bin" -OutFile "ggml-large-v3-turbo.bin"

Network accessline 117

Если NPU/GPU не появились — обнови [Intel Graphics Driver](https://www.intel.com/content/www/us/en/download-center/home.html) (нужна версия 32.0.x или новее, релизы 2024+).

Network accessline 139

CUDA Toolkit не стоит или PATH не настроен. Решение: 1. Поставить CUDA Toolkit с https://developer.nvidia.com/cuda-downloads

Network accessline 140

1. Поставить CUDA Toolkit с https://developer.nvidia.com/cuda-downloads

Network accessline 146

cuDNN не установлен. Скачай с https://developer.nvidia.com/cudnn (нужен NVIDIA Developer аккаунт, бесплатный) и распакуй в `C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.x\`.

Network accessline 158

pip install torch torchaudio --index-url https://download.pytorch.org/whl/cu121

Access to secretsline 168
```
[Credential-related source omitted]
```

`docs/known-issues.md`

Network accessline 28

pip uninstall torch torchaudio # Для CUDA 12.1: pip install torch torchaudio --index-url https://download.pytorch.org/whl/cu121

Network accessline 29

# Для CUDA 12.1: pip install torch torchaudio --index-url https://download.pytorch.org/whl/cu121 # Для CUDA 11.8:

Network accessline 30

pip install torch torchaudio --index-url https://download.pytorch.org/whl/cu121

Network accessline 32

pip install torch torchaudio --index-url https://download.pytorch.org/whl/cu118

Network accessline 37

CUDA Toolkit не установлен. https://developer.nvidia.com/cuda-downloads — выбери Windows → exe (network).

Dangerous shell operationline 43

sudo apt install libcudnn8 libcudnn8-dev

Network accessline 189

Не принял условия использования. На каждой из этих страниц нажми "Agree": - https://huggingface.co/pyannote/segmentation-3.0

Network accessline 190

- https://huggingface.co/pyannote/segmentation-3.0

Network accessline 191

- https://huggingface.co/pyannote/speaker-diarization-3.1

Network accessline 193

Затем создай токен: https://huggingface.co/settings/tokens

Access to secretsline 197
```
[Credential-related source omitted]
```
Access to secretsline 233
```
[Credential-related source omitted]
```
Access to secretsline 234
```
[Credential-related source omitted]
```

Network accessline 279

- 50-200 часов аудио своего домена - 1-2 эпохи на A100 (несколько часов) - См. https://huggingface.co/blog/fine-tune-whisper

Network accessline 281

- См. https://huggingface.co/blog/fine-tune-whisper

`docs/subtitle-baker.md`

Dangerous shell operationline 157

sudo apt install fonts-montserrat fonts-inter fonts-roboto

Network accessline 177

# 1. Скачал видео из URL через yt-dlp wrapper python -m examples.from_url "https://www.tiktok.com/@user/video/123" --output ./content/ --keep-audio

Network accessline 178

python -m examples.from_url "https://www.tiktok.com/@user/video/123" --output ./content/ --keep-audio

`docs/voice-dictation.md`

Dangerous shell operationline 127
```
sudo usermod -aG input $USER
```

Network accessline 196

<?xml version="1.0" encoding="UTF-8"?> <!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN"         "http://www.apple.com/DTDs/PropertyList-1.0.dtd">

Network accessline 197

<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN"         "http://www.apple.com/DTDs/PropertyList-1.0.dtd"> <plist version="1.0">

Network accessline 198

"http://www.apple.com/DTDs/PropertyList-1.0.dtd">

`examples/init.py`

Executable scriptline 1
```
Executable file
```

`examples/bake_subs.py`

Executable scriptline 1
```
Executable file
```

`examples/batch_folder.py`

Executable scriptline 1
```
Executable file
```

`examples/common.py`

Executable scriptline 1
```
Executable file
```
Access to secretsline 51
```
[Credential-related source omitted]
```
Access to secretsline 78
```
[Credential-related source omitted]
```
Access to secretsline 79
```
[Credential-related source omitted]
```
Access to secretsline 140
```
[Credential-related source omitted]
```
Access to secretsline 198
```
[Credential-related source omitted]
```
Access to secretsline 202
```
[Credential-related source omitted]
```
Access to secretsline 203
```
[Credential-related source omitted]
```
Access to secretsline 204
```
[Credential-related source omitted]
```
Access to secretsline 373
```
[Credential-related source omitted]
```
Access to secretsline 374
```
[Credential-related source omitted]
```
Access to secretsline 407
```
[Credential-related source omitted]
```

`examples/from_url.py`

Executable scriptline 1
```
Executable file
```

Network accessline 4

Использование:     python -m examples.from_url "https://www.tiktok.com/@user/video/123..."

Network accessline 5

python -m examples.from_url "https://www.tiktok.com/@user/video/123..."

Network accessline 6

python -m examples.from_url "https://www.youtube.com/watch?v=..." --language en

Network accessline 7

python -m examples.from_url "https://www.instagram.com/reel/..." --output ./out/

`examples/podcast_diarize.py`

Executable scriptline 1
```
Executable file
```
Access to secretsline 10
```
[Credential-related source omitted]
```

Network accessline 12

Также требуется согласие на условия использования pyannote-моделей:     https://huggingface.co/pyannote/speaker-diarization-3.1

Network accessline 13

https://huggingface.co/pyannote/speaker-diarization-3.1

Network accessline 14

https://huggingface.co/pyannote/segmentation-3.0

Access to secretsline 40
```
[Credential-related source omitted]
```
Network accessline 42
```
[Credential-related source omitted]
```
Network accessline 43
```
[Credential-related source omitted]
```
Access to secretsline 43
```
[Credential-related source omitted]
```

Network accessline 44

"Получи бесплатно на https://huggingface.co/settings/tokens\n"

Network accessline 45

"И прими условия pyannote: https://huggingface.co/pyannote/speaker-diarization-3.1",

Access to secretsline 80
```
[Credential-related source omitted]
```

`examples/transcribe_one.py`

Executable scriptline 1
```
Executable file
```

`examples/voice_dictation.py`

Executable scriptline 1
```
Executable file
```
Access to secretsline 75
```
[Credential-related source omitted]
```
Access to secretsline 1200
```
[Credential-related source omitted]
```
Access to secretsline 1201
```
[Credential-related source omitted]
```
Access to secretsline 1202
```
[Credential-related source omitted]
```
Access to secretsline 1206
```
[Credential-related source omitted]
```
Access to secretsline 1208
```
[Credential-related source omitted]
```

Dangerous shell operationline 1224

print("  Linux:  sudo apt install python3-tk", file=sys.stderr)

`launcher/voice_dictation_silent.vbs`

Executable scriptline 1
```
Executable file
```

`methodology/quality-vs-speed.md`

Network accessline 111

pip install noisereduce    # или используй RNNoise (https://github.com/xiph/rnnoise)

Network accessline 112

# или используй RNNoise (https://github.com/xiph/rnnoise)

`models/README.md`

Network accessline 90

WhisperModel("large-v3-turbo", ...) # → качает из https://huggingface.co/Systran/faster-whisper-large-v3-turbo

Network accessline 91

# → качает из https://huggingface.co/Systran/faster-whisper-large-v3-turbo

Network accessline 97

bash ./models/download-ggml-model.sh large-v3-turbo # или прямой URL: https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-large-v3-turbo.bin

Network accessline 98

# или прямой URL: https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-large-v3-turbo.bin

`scripts/build_icon.py`

Executable scriptline 1
```
Executable file
```

`scripts/convert_openvino.py`

Executable scriptline 1
```
Executable file
```

`scripts/cursor_indicator.py`

Executable scriptline 1
```
Executable file
```

`scripts/detect_env.py`

Executable scriptline 1
```
#!/usr/bin/env python3
```
Access to secretsline 33
```
[Credential-related source omitted]
```

Network accessline 494

if rec.backend == "mlx-whisper":         rec.install_commands = [             "# 1) Поставь Homebrew если ещё нет: https://brew.sh",

Network accessline 495

rec.install_commands = [             "# 1) Поставь Homebrew если ещё нет: https://brew.sh",             "# 2) ffmpeg для извлечения аудио из видео:",

Network accessline 496

"# 1) Поставь Homebrew если ещё нет: https://brew.sh",

Network accessline 508

rec.install_commands = [             "# 1) Убедись что есть CUDA Toolkit 12.x: nvidia-smi показывает Driver, nvcc показывает версию.",             "# Если нет — поставь: https://developer.nvidia.com/cuda-downloads",

Network accessline 509

"# 1) Убедись что есть CUDA Toolkit 12.x: nvidia-smi показывает Driver, nvcc показывает версию.",             "# Если нет — поставь: https://developer.nvidia.com/cuda-downloads",             "# 2) Создай venv:",

Network accessline 510

"# Если нет — поставь: https://developer.nvidia.com/cuda-downloads",

Dangerous shell operationline 517

"# Linux: sudo apt install ffmpeg  |  Windows: winget install ffmpeg  |  Mac: brew install ffmpeg",

Network accessline 567

f"# Скачай модель large-v3-turbo (~1.5 GB):",                 "mkdir -p models && cd models",                 f"curl -LO https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-{rec.model}.bin",

Network accessline 568

"mkdir -p models && cd models",                 f"curl -LO https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-{rec.model}.bin",                 "cd ..",

Network accessline 569

f"curl -LO https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-{rec.model}.bin",

Network accessline 573

rec.install_commands = [                 "sudo apt update && sudo apt install -y build-essential cmake ffmpeg",                 "git clone https://github.com/ggerganov/whisper.cpp.git",

Dangerous shell operationline 574

"sudo apt update && sudo apt install -y build-essential cmake ffmpeg",

Network accessline 574

"sudo apt update && sudo apt install -y build-essential cmake ffmpeg",                 "git clone https://github.com/ggerganov/whisper.cpp.git",                 "cd whisper.cpp && cmake -B build && cmake --build build --config Release",

Network accessline 575

"git clone https://github.com/ggerganov/whisper.cpp.git",                 "cd whisper.cpp && cmake -B build && cmake --build build --config Release",                 f"bash ./models/download-ggml-model.sh {rec.model}",

Dangerous shell operationline 605

f"brew install python@3.12 (Mac) | sudo apt install python3.12 (Linux)"

`tools/install_autostart.ps1`

Executable scriptline 1
```
Executable file
```

`tools/install_shortcut.ps1`

Executable scriptline 1
```
Executable file
```

`tools/uninstall_autostart.ps1`

Executable scriptline 1
```
Executable file
```

`wizard.py`

Executable scriptline 1
```
#!/usr/bin/env python3
```

Network accessline 233

else:                 print(red("\n❌ Homebrew не найден."))                 print(yellow("Установи Homebrew: https://brew.sh"))

Network accessline 234

print(red("\n❌ Homebrew не найден."))                 print(yellow("Установи Homebrew: https://brew.sh"))                 print(yellow("Потом запусти wizard ещё раз."))

Network accessline 235

print(yellow("Установи Homebrew: https://brew.sh"))

Network accessline 249

run(["winget", "install", "Gyan.FFmpeg"], check=False)             else:                 print(yellow("Поставь ffmpeg вручную: https://ffmpeg.org/download.html"))

Network accessline 251

print(yellow("Поставь ffmpeg вручную: https://ffmpeg.org/download.html"))

Network accessline 258

print(yellow("Сборка whisper.cpp из исходников. Это может занять 2-5 минут."))             if ask_yn("Продолжить?", default=True):                 run("git clone https://github.com/ggerganov/whisper.cpp.git", shell=True, check=False)

Network accessline 259

if ask_yn("Продолжить?", default=True):                 run("git clone https://github.com/ggerganov/whisper.cpp.git", shell=True, check=False)                 run(["cmake", "-B", "whisper.cpp/build", "-S", "whisper.cpp"], check=False)

Network accessline 260

run("git clone https://github.com/ggerganov/whisper.cpp.git", shell=True, check=False)

Access to secretsline 272
```
[Credential-related source omitted]
```

Network accessline 324

print("Для разметки спикеров (diarization) нужен бесплатный Hugging Face token.")         print()         print("1. Регистрируйся: https://huggingface.co (если ещё не)")

Network accessline 325

print()         print("1. Регистрируйся: https://huggingface.co (если ещё не)")         print("2. Создай токен:  https://huggingface.co/settings/tokens (тип Read)")

Network accessline 326

print("1. Регистрируйся: https://huggingface.co (если ещё не)")

Network accessline 327

print("2. Создай токен:  https://huggingface.co/settings/tokens (тип Read)")

Network accessline 328

print("3. Прими условия использования двух моделей:")         print("   • https://huggingface.co/pyannote/speaker-diarization-3.1")         print("   • https://huggingface.co/pyannote/segmentation-3.0")

Network accessline 329

print("   • https://huggingface.co/pyannote/speaker-diarization-3.1")

Network accessline 330

print("   • https://huggingface.co/pyannote/segmentation-3.0")

Network accessline 357

print(green(f"  python -m examples.transcribe_one input.mp3 --language ru"))         print(green(f"  python -m examples.batch_folder ./videos/ --language ru"))         print(green(f"  python -m examples.from_url 'https://www.tiktok.com/@use

Network accessline 359

print(green(f"  python -m examples.from_url 'https://www.tiktok.com/@user/video/123'"))

Network accessline 386

try:             run(["curl", "-fsSL", "-o", str(sample),                  "https://github.com/ggerganov/whisper.cpp/raw/master/samples/jfk.wav"], check=False)

Network accessline 387

run(["curl", "-fsSL", "-o", str(sample),                  "https://github.com/ggerganov/whisper.cpp/raw/master/samples/jfk.wav"], check=False)         except Exception as e:

Network accessline 388

"https://github.com/ggerganov/whisper.cpp/raw/master/samples/jfk.wav"], check=False)

How the signals fit together

Facts from a static read of the files at the pinned commit. They describe what the code does, not why.

Declared and executed: 4 facts

Contacts brew.sh, which SKILL.md, the README and the plugin manifest do not mentionscripts/detect_env.py:496High confidence
Contacts developer.nvidia.com, which SKILL.md, the README and the plugin manifest do not mentionscripts/detect_env.py:510High confidence
Contacts brew.sh, which SKILL.md, the README and the plugin manifest do not mentionwizard.py:235High confidence
Contacts ffmpeg.org, which SKILL.md, the README and the plugin manifest do not mentionwizard.py:251High confidence

Signal confidence: 250 facts

Network access in code of documentationREADME.md:16Low confidence
Network access in prose of documentationREADME.md:49Low confidence
Network access in code of documentationREADME.md:69Low confidence
Network access in code of documentationREADME.md:70Low confidence
Network access in code of documentationREADME.md:234Low confidence
Network access in code of documentationREADME.md:235Low confidence
Network access in code of documentationREADME.md:237Low confidence
Network access in code of documentationREADME.md:238Low confidence
Network access in code of documentationREADME.md:240Low confidence
Network access in code of documentationREADME.md:241Low confidence
Network access in code of documentationREADME.md:247Low confidence
Network access in code of documentationREADME.md:248Low confidence
Access to secrets in code of documentationREADME.md:249Low confidence
Dangerous shell operation in prose of documentationREADME.md:524Low confidence
Network access in code of the SKILL.mdSKILL.md:80High confidence
Network access in code of the SKILL.mdSKILL.md:81High confidence
Network access in prose of documentationbackends/faster-whisper.md:3Low confidence
Network access in prose of documentationbackends/faster-whisper.md:4Low confidence
Network access in code of documentationbackends/faster-whisper.md:29Low confidence
Network access in code of documentationbackends/faster-whisper.md:30Low confidence
Dangerous shell operation in code of documentationbackends/faster-whisper.md:39Low confidence
Network access in code of documentationbackends/faster-whisper.md:45Low confidence
Network access in code of documentationbackends/faster-whisper.md:76Low confidence
Network access in code of documentationbackends/faster-whisper.md:77Low confidence
Network access in code of documentationbackends/faster-whisper.md:78Low confidence
Network access in code of documentationbackends/faster-whisper.md:202Low confidence
Dangerous shell operation in code of documentationbackends/faster-whisper.md:203Low confidence
Network access in code of documentationbackends/faster-whisper.md:204Low confidence
Network access in prose of documentationbackends/faster-whisper.md:209Low confidence
Network access in prose of documentationbackends/mlx-whisper.md:3Low confidence
Network access in prose of documentationbackends/mlx-whisper.md:4Low confidence
Network access in prose of documentationbackends/openvino.md:3Low confidence
Network access in prose of documentationbackends/openvino.md:4Low confidence
Network access in prose of documentationbackends/openvino.md:44Low confidence
Dangerous shell operation in code of documentationbackends/openvino.md:50Low confidence
Network access in code of documentationbackends/openvino.md:50Low confidence
Network access in code of documentationbackends/openvino.md:52Low confidence
Access to secrets in code of documentation: a variable, placeholder or call, not a stored valuebackends/openvino.md:115Low confidence
Access to secrets in code of documentation: a variable, placeholder or call, not a stored valuebackends/openvino.md:116Low confidence
Network access in prose of documentationbackends/whisper-cpp.md:3Low confidence
Network access in code of documentationbackends/whisper-cpp.md:28Low confidence
Network access in code of documentationbackends/whisper-cpp.md:30Low confidence
Network access in code of documentationbackends/whisper-cpp.md:36Low confidence
Dangerous shell operation in code of documentationbackends/whisper-cpp.md:37Low confidence
Network access in code of documentationbackends/whisper-cpp.md:37Low confidence
Network access in code of documentationbackends/whisper-cpp.md:38Low confidence
Network access in code of documentationbackends/whisper-cpp.md:59Low confidence
Network access in code of documentationbackends/whisper-cpp.md:60Low confidence
Network access in prose of documentationbackends/whisperx.md:3Low confidence
Network access in prose of documentationbackends/whisperx.md:4Low confidence

and 200 more

Works with

Agent compatibility, from the files in this repository
Agent	Status	Evidence	Install to
Claude Code	Compatible	Standard SKILL.md SKILL.md	`.claude/skills/<skill-name>/`
Codex	Compatible	Standard SKILL.md SKILL.md	`.agents/skills/<skill-name>/`
Cursor	Compatible	Standard SKILL.md SKILL.md	`.cursor/skills/<skill-name>/`
Gemini CLI	Compatible	Standard SKILL.md SKILL.md	`.gemini/skills/<skill-name>/`
GitHub Copilot	Compatible	Standard SKILL.md SKILL.md	`.github/skills/<skill-name>/`
OpenCode	Compatible	Standard SKILL.md SKILL.md	`.opencode/skills/<skill-name>/`

Native The repository ships a manifest, folder or frontmatter field for this agent.
Compatible A standard SKILL.md that this agent can read.
Unknown No evidence either way.

Derived automatically from paths, manifests and frontmatter. Not tested on each agent.

Mobiss11/Whisper-Skill

History

SKILL.md

Share this page

Heat

Stars over time

Individual skills in this repository

How it works

whisper-skill

Frontmatter

SKILL.md

Reference documents

The playbook

Scripts executed separately

Files

Skills included

Repository facts

Before you install

README.md

SKILL.md

backends/faster-whisper.md

backends/mlx-whisper.md

backends/openvino.md

backends/whisper-cpp.md

backends/whisperx.md

docs/cost-comparison.md

docs/diarization.md

docs/installation-linux.md

docs/installation-mac.md

docs/installation-windows.md

docs/known-issues.md

docs/subtitle-baker.md

docs/voice-dictation.md

examples/__init__.py

examples/bake_subs.py

examples/batch_folder.py

examples/common.py

examples/from_url.py

examples/podcast_diarize.py

examples/transcribe_one.py

examples/voice_dictation.py

launcher/voice_dictation_silent.vbs

methodology/quality-vs-speed.md

models/README.md

scripts/build_icon.py

scripts/convert_openvino.py

scripts/cursor_indicator.py

scripts/detect_env.py

tools/install_autostart.ps1

tools/install_shortcut.ps1

tools/uninstall_autostart.ps1

wizard.py

How the signals fit together

Works with

You might also use

AugustusW/audio-tldr-skill

silverstein/minutes

seacen/meetcap

Backtthefuture/video-transcript

drpwchen/lecture-to-notes

heyuxuan0209/read-anything

`SKILL.md`

`README.md`

`SKILL.md`

`backends/faster-whisper.md`

`backends/mlx-whisper.md`

`backends/openvino.md`

`backends/whisper-cpp.md`

`backends/whisperx.md`

`docs/cost-comparison.md`

`docs/diarization.md`

`docs/installation-linux.md`

`docs/installation-mac.md`

`docs/installation-windows.md`

`docs/known-issues.md`

`docs/subtitle-baker.md`

`docs/voice-dictation.md`

`examples/init.py`

`examples/bake_subs.py`

`examples/batch_folder.py`

`examples/common.py`

`examples/from_url.py`

`examples/podcast_diarize.py`

`examples/transcribe_one.py`

`examples/voice_dictation.py`

`launcher/voice_dictation_silent.vbs`

`methodology/quality-vs-speed.md`

`models/README.md`

`scripts/build_icon.py`

`scripts/convert_openvino.py`

`scripts/cursor_indicator.py`

`scripts/detect_env.py`

`tools/install_autostart.ps1`

`tools/install_shortcut.ps1`

`tools/uninstall_autostart.ps1`

`wizard.py`