be9f14ce34
A small helpdesk system: an office Pixel running a patched GrapheneOS Dialer answers technician calls, records both call legs as separate channels, and a Ruby backend transcribes them through Whisper and files an AI summary against the caller. Squashed to a single commit for sharing. No credentials are included; secrets live outside the repo in /etc/helpdesk/env on the server or a gitignored .claude/env.local locally. See .claude/env.local.example for the shape. Start at README.md, then docs/architecture.md.
20 lines
636 B
Text
20 lines
636 B
Text
# Pinned to WhisperX 3.8.5's known-good dependency tree (April 2026).
|
|
# CPU-only PyTorch — switch to a CUDA index URL if you have an NVIDIA GPU.
|
|
--extra-index-url https://download.pytorch.org/whl/cpu
|
|
|
|
# Top-level entrypoints (pinned hard).
|
|
whisperx==3.8.5
|
|
torch==2.8.0
|
|
torchaudio==2.8.0
|
|
torchvision==0.23.0
|
|
|
|
# Diarization + ASR backends (pinned to versions WhisperX 3.8.5 expects).
|
|
pyannote.audio==4.0.0
|
|
faster-whisper==1.2.0
|
|
ctranslate2==4.5.0
|
|
|
|
# HuggingFace stack — let WhisperX resolve transformers, but cap hub <1.0.0.
|
|
huggingface-hub<1.0.0
|
|
|
|
# Progress bars (also a transitive of WhisperX, pinned for consistency).
|
|
tqdm>=4.66.0
|