add wyoming-chatterbox addon

README.md

@@ -16,3 +16,6 @@ find isolated builders with aligned values. auto-discovers humans on github, mas
 
 ### ankerctl
 control and monitor ankermake 3d printers via ankerctl.
+
+### wyoming-chatterbox
+wyoming protocol server for chatterbox tts with voice cloning. clone any voice with a 10-30 second sample. requires nvidia gpu.

wyoming-chatterbox/Dockerfile

@@ -0,0 +1,18 @@
+ARG BUILD_FROM
+FROM nvidia/cuda:12.1.0-runtime-ubuntu22.04
+
+# install python
+RUN apt-get update && apt-get install -y \
+    python3 python3-pip python3-venv git \
+    && rm -rf /var/lib/apt/lists/*
+
+# install wyoming-chatterbox
+RUN pip3 install --no-cache-dir wyoming-chatterbox
+
+# copy run script
+COPY run.sh /
+RUN chmod a+x /run.sh
+
+WORKDIR /data
+
+CMD ["/run.sh"]

wyoming-chatterbox/README.md

@@ -0,0 +1,32 @@
+# wyoming-chatterbox addon
+
+wyoming protocol server for chatterbox tts with voice cloning. clone any voice with a 10-30 second sample.
+
+## requirements
+
+- nvidia gpu with 4gb+ vram
+- gpu passthrough configured in your HA host
+
+## configuration
+
+| option | default | description |
+|--------|---------|-------------|
+| `voice_ref` | required | path to voice reference wav (place in /share/) |
+| `volume_boost` | 3.0 | output volume multiplier |
+| `device` | cuda | torch device (cuda or cpu) |
+| `debug` | false | enable debug logging |
+
+## setup
+
+1. place your voice reference wav in `/share/voice_reference.wav`
+2. configure the addon with the path
+3. start the addon
+4. add wyoming integration in HA pointing to port 10201
+
+## voice reference tips
+
+for best results:
+- 10-30 seconds of clean speech
+- no background music or noise
+- consistent speaking style
+- wav format (any sample rate)

wyoming-chatterbox/build.yaml

@@ -0,0 +1,7 @@
+build_from:
+  amd64: nvidia/cuda:12.1.0-runtime-ubuntu22.04
+labels:
+  org.opencontainers.image.title: "wyoming-chatterbox"
+  org.opencontainers.image.description: "wyoming protocol server for chatterbox tts with voice cloning"
+  org.opencontainers.image.source: "https://github.com/sudoxnym/wyoming-chatterbox"
+  org.opencontainers.image.licenses: "MIT"

wyoming-chatterbox/config.yaml

@@ -0,0 +1,27 @@
+name: wyoming-chatterbox
+version: "1.0.0"
+slug: wyoming-chatterbox
+description: "wyoming protocol server for chatterbox tts with voice cloning. clone any voice with a 10-30 second sample."
+url: "https://github.com/sudoxnym/wyoming-chatterbox"
+arch:
+  - amd64
+startup: application
+boot: auto
+ports:
+  10201/tcp: 10201
+ports_description:
+  10201/tcp: "wyoming protocol"
+map:
+  - share:rw
+options:
+  voice_ref: "/share/voice_reference.wav"
+  volume_boost: 3.0
+  device: "cuda"
+  debug: false
+schema:
+  voice_ref: str
+  volume_boost: float?
+  device: list(cuda|cpu)?
+  debug: bool?
+image: sudoxreboot/wyoming-chatterbox-addon-{arch}
+full_access: true

wyoming-chatterbox/run.sh

@@ -0,0 +1,23 @@
+#!/usr/bin/env bash
+set -e
+
+CONFIG_PATH=/data/options.json
+
+VOICE_REF=$(jq -r '.voice_ref' $CONFIG_PATH)
+VOLUME_BOOST=$(jq -r '.volume_boost // 3.0' $CONFIG_PATH)
+DEVICE=$(jq -r '.device // "cuda"' $CONFIG_PATH)
+DEBUG=$(jq -r '.debug // false' $CONFIG_PATH)
+
+echo "starting wyoming-chatterbox..."
+echo "voice_ref: ${VOICE_REF}"
+echo "device: ${DEVICE}"
+
+ARGS="--uri tcp://0.0.0.0:10201 --voice-ref ${VOICE_REF} --volume-boost ${VOLUME_BOOST} --device ${DEVICE}"
+
+if [ "${DEBUG}" = "true" ]; then
+    ARGS="${ARGS} --debug"
+fi
+
+export PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True
+
+exec wyoming-chatterbox ${ARGS}