fegger
76611af7e0
- whisper-server stack: second container (port 8086) running the English-trained small.en-tdrz model with -tdrz; image patched (speaker-turn.patch) to expose speaker_turn_next per segment in verbose_json like the cli example does - core/whisper Diarization: merges the tdrz pass's TURN TIMES onto the quality transcript as alternating 'Sprecher 1/2:' labels, splitting segments when a turn falls inside them; no turns detected = no labels (never mislabels); 6 unit tests - RemoteWhisperEngine gains a diarize flag (sends tinydiarize=true, parses speaker_turn_next); WhisperEngine.Segment carries the flag - RecorderService: optional second pass on the diarize server after the final pass; failures keep the unlabeled transcript - Settings: Diarize server URL (persisted; empty disables) - validated infrastructure locally: patched image builds, tdrz model downloads from akashmjn/tinydiarize-whisper.cpp, speaker_turn_next present in responses; synthetic espeak audio does not trigger the model's turn tokens — real two-person speech needed for the end-to-end check
87 lines
2.7 KiB
YAML
87 lines
2.7 KiB
YAML
# whisper.cpp inference servers for meetrec, with Vulkan GPU support
|
|
# (AMD Radeon AI PRO R9700), reachable over Tailscale.
|
|
#
|
|
# RENDER_GID=$(getent group render | cut -d: -f3) docker compose up -d --build
|
|
#
|
|
# Two services:
|
|
# whisper-server port 8085 — large-v3, quality transcripts
|
|
# whisper-server-tdrz port 8086 — small.en-tdrz, tinydiarize speaker
|
|
# turns (English-trained, 2-speaker, best-effort)
|
|
#
|
|
# The image is patched to expose `speaker_turn_next` per segment in
|
|
# verbose_json (speaker-turn.patch); clients merge the turn times onto
|
|
# the better transcript as "Sprecher 1/2" labels.
|
|
|
|
services:
|
|
whisper-server:
|
|
build: .
|
|
image: meetrec-whisper-server:latest
|
|
container_name: whisper-server
|
|
restart: unless-stopped
|
|
|
|
environment:
|
|
# tiny | base | small | medium | large-v3 (downloads on first start)
|
|
MODEL: large-v3
|
|
THREADS: 8
|
|
HOST: 0.0.0.0 # inside the container; the host bind is below
|
|
PORT: 8085
|
|
# NO_GPU: 1 # force CPU if the GPU ever misbehaves
|
|
|
|
volumes:
|
|
- ./models:/models # keep downloads across container rebuilds
|
|
|
|
ports:
|
|
- "100.103.83.12:8085:8085" # tailscale-only; see note above
|
|
|
|
# GPU passthrough: the render group must match the host's GID, hence
|
|
# the RENDER_GID env var in the comment at the top of this file.
|
|
devices:
|
|
- /dev/dri:/dev/dri
|
|
group_add:
|
|
- video
|
|
- "${RENDER_GID:-110}"
|
|
|
|
healthcheck:
|
|
# any HTTP response counts as healthy (the server has no /health)
|
|
test: ["CMD-SHELL", "curl -s -o /dev/null http://localhost:8085/ || exit 1"]
|
|
interval: 30s
|
|
timeout: 5s
|
|
retries: 3
|
|
start_period: 60s
|
|
|
|
whisper-server-tdrz:
|
|
# tinydiarize variant: English-only small.en-tdrz, used ONLY for its
|
|
# speaker-turn timestamps (meetrec merges them onto the better
|
|
# transcript from the main server). 2-speaker detection, best-effort
|
|
# on non-English audio.
|
|
build: .
|
|
image: meetrec-whisper-server:latest
|
|
container_name: whisper-server-tdrz
|
|
restart: unless-stopped
|
|
|
|
environment:
|
|
MODEL: small.en-tdrz
|
|
MODEL_URL: https://huggingface.co/akashmjn/tinydiarize-whisper.cpp/resolve/main/ggml-small.en-tdrz.bin
|
|
THREADS: 8
|
|
HOST: 0.0.0.0
|
|
PORT: 8086
|
|
TDRZ: 1
|
|
|
|
volumes:
|
|
- ./models:/models # shared with the main server (distinct files)
|
|
|
|
ports:
|
|
- "100.103.83.12:8086:8086" # tailscale-only
|
|
|
|
devices:
|
|
- /dev/dri:/dev/dri
|
|
group_add:
|
|
- video
|
|
- "${RENDER_GID:-110}"
|
|
|
|
healthcheck:
|
|
test: ["CMD-SHELL", "curl -s -o /dev/null http://localhost:8086/ || exit 1"]
|
|
interval: 30s
|
|
timeout: 5s
|
|
retries: 3
|
|
start_period: 60s |