Support streaming & local mode, install piper from pypi

Dockerfile

@@ -12,14 +12,9 @@ LABEL maintainer="thespad"
 ENV DEBIAN_FRONTEND="noninteractive" \
   TMPDIR="/run/piper-temp"
 
-RUN --mount=type=bind,source=/patch,target=/patch \
+RUN \
   apt-get update && \
   apt-get install -y --no-install-recommends \
-    build-essential \
-    cmake \
-    git \
-    pkg-config \
-    patch \
     python3-venv && \
   if [ -z ${PIPER_VERSION+x} ]; then \
     PIPER_VERSION=$(curl -sL  https://pypi.python.org/pypi/wyoming-piper/json |jq -r '. | .info.version'); \
@@ -36,34 +31,10 @@ RUN --mount=type=bind,source=/patch,target=/patch \
     nvidia-cufft-cu12 \
     nvidia-curand-cu12 \
     onnxruntime-gpu \
+    piper-tts \
     "wyoming-piper==${PIPER_VERSION}" && \
-  if [ -z ${PIPER_BIN_VERSION+x} ]; then \
-    PIPER_BIN_VERSION=$(curl -sL "https://api.github.com/repos/rhasspy/piper/commits/master" \
-    | jq -r .sha); \
-  fi && \
-  mkdir -p /tmp/piper-build /usr/share/piper && \
-  curl -sL -o  \
-    /tmp/piper.tar.gz -L \
-    "https://github.com/rhasspy/piper/archive/${PIPER_BIN_VERSION}.tar.gz" && \
-  tar xzf \
-    /tmp/piper.tar.gz -C \
-    /tmp/piper-build --strip-components=1 && \
-  cd /tmp/piper-build && \
-  cmake -Bbuild -DCMAKE_INSTALL_PREFIX=install && \
-  cmake --build build --config Release && \
-  cmake --install build && \
-  cp -dR /tmp/piper-build/install/* /usr/share/piper && \
-  patch /lsiopy/lib/python3.12/site-packages/wyoming_piper/__main__.py < /patch/__main__.patch && \
-  patch /lsiopy/lib/python3.12/site-packages/wyoming_piper/process.py < /patch/process.patch && \
   printf "Linuxserver.io version: ${VERSION}\nBuild-date: ${BUILD_DATE}" > /build_version && \
   echo "**** cleanup ****" && \
-  apt-get purge -y \
-    build-essential \
-    cmake \
-    git \
-    pkg-config && \
-  apt-get autoremove -y && \
-  apt-get clean -y && \
   rm -rf \
     /var/lib/apt/lists/* \
     /tmp/*

Jenkinsfile

@@ -29,10 +29,10 @@ pipeline {
     PR_DOCKERHUB_IMAGE = 'lspipepr/piper'
     DIST_IMAGE = 'ubuntu'
     MULTIARCH='false'
-    CI='false'
+    CI='true'
     CI_WEB='false'
     CI_DELAY='120'
-    CI_DOCKERENV=''
+    CI_DOCKERENV='PIPER_VOICE=en_US-lessac-medium'
     CI_AUTH=''
     CI_WEBPATH=''
   }

README.md

@@ -35,6 +35,7 @@ Find us at:
 [![Docker Pulls](https://img.shields.io/docker/pulls/linuxserver/piper.svg?color=94398d&labelColor=555555&logoColor=ffffff&style=for-the-badge&label=pulls&logo=docker)](https://hub.docker.com/r/linuxserver/piper)
 [![Docker Stars](https://img.shields.io/docker/stars/linuxserver/piper.svg?color=94398d&labelColor=555555&logoColor=ffffff&style=for-the-badge&label=stars&logo=docker)](https://hub.docker.com/r/linuxserver/piper)
 [![Jenkins Build](https://img.shields.io/jenkins/build?labelColor=555555&logoColor=ffffff&style=for-the-badge&jobUrl=https%3A%2F%2Fci.linuxserver.io%2Fjob%2FDocker-Pipeline-Builders%2Fjob%2Fdocker-piper%2Fjob%2Fgpu%2F&logo=jenkins)](https://ci.linuxserver.io/job/Docker-Pipeline-Builders/job/docker-piper/job/gpu/)
+[![LSIO CI](https://img.shields.io/badge/dynamic/yaml?color=94398d&labelColor=555555&logoColor=ffffff&style=for-the-badge&label=CI&query=CI&url=https%3A%2F%2Fci-tests.linuxserver.io%2Flinuxserver%2Fpiper%2Fgpu%2Fci-status.yml)](https://ci-tests.linuxserver.io/linuxserver/piper/gpu/index.html)
 
 [Piper](https://github.com/rhasspy/piper/) is a fast, local neural text to speech system that sounds great and is optimized for the Raspberry Pi 4. This container provides a Wyoming protocol server for Piper.
 
@@ -94,11 +95,12 @@ services:
       - PGID=1000
       - TZ=Etc/UTC
       - PIPER_VOICE=en_US-lessac-medium
+      - LOCAL_ONLY= #optional
       - PIPER_LENGTH=1.0 #optional
       - PIPER_NOISE=0.667 #optional
       - PIPER_NOISEW=0.333 #optional
       - PIPER_SPEAKER=0 #optional
-      - PIPER_PROCS=1 #optional
+      - STREAMING= #optional
     volumes:
       - /path/to/piper/data:/config
     ports:
@@ -115,11 +117,12 @@ docker run -d \
   -e PGID=1000 \
   -e TZ=Etc/UTC \
   -e PIPER_VOICE=en_US-lessac-medium \
+  -e LOCAL_ONLY= `#optional` \
   -e PIPER_LENGTH=1.0 `#optional` \
   -e PIPER_NOISE=0.667 `#optional` \
   -e PIPER_NOISEW=0.333 `#optional` \
   -e PIPER_SPEAKER=0 `#optional` \
-  -e PIPER_PROCS=1 `#optional` \
+  -e STREAMING= `#optional` \
   -p 10200:10200 \
   -v /path/to/piper/data:/config \
   --restart unless-stopped \
@@ -137,11 +140,12 @@ Containers are configured using parameters passed at runtime (such as those abov
 | `-e PGID=1000` | for GroupID - see below for explanation |
 | `-e TZ=Etc/UTC` | specify a timezone to use, see this [list](https://en.wikipedia.org/wiki/List_of_tz_database_time_zones#List). |
 | `-e PIPER_VOICE=en_US-lessac-medium` | The [Piper voice](https://huggingface.co/rhasspy/piper-voices/tree/main) to use, in the format `<language>-<name>-<quality>` |
+| `-e LOCAL_ONLY=` | If set to `true`, or any other value, the container will not attempt to download models from HuggingFace and will only use locally-provided models. |
 | `-e PIPER_LENGTH=1.0` | Voice speaking rate, 1.0 is default with < 1.0 being faster and > 1.0 being slower. |
 | `-e PIPER_NOISE=0.667` | Controls the variability of the voice by adding noise. Values above 1 will start to degrade audio. |
 | `-e PIPER_NOISEW=0.333` | Controls the variability of speaking cadence. Values above 1 produce extreme stutters and pauses. |
 | `-e PIPER_SPEAKER=0` | Speaker number to use if the voice supports multiple speakers. |
-| `-e PIPER_PROCS=1` | Number of Piper processes to run simultaneously. |
+| `-e STREAMING=` | Setting to `true`, or any other value, will enable support for streaming audio on sentence boundaries. |
 | `-v /config` | Local path for piper config files. |
 | `--read-only=true` | Run container with a read-only filesystem. Please [read the docs](https://docs.linuxserver.io/misc/read-only/). |
 
@@ -307,6 +311,7 @@ Once registered you can define the dockerfile to use with `-f Dockerfile.aarch64
 
 ## Versions
 
+* **29.08.25:** - Add streaming support & support for local only mode.
 * **26.11.24:** - Add GPU branch.
 * **18.07.24:** - Rebase to Ubuntu Noble.
 * **25.11.23:** - Initial Release.

jenkins-vars.yml

@@ -17,9 +17,9 @@ repo_vars:
   - PR_DOCKERHUB_IMAGE = 'lspipepr/piper'
   - DIST_IMAGE = 'ubuntu'
   - MULTIARCH='false'
-  - CI='false'
+  - CI='true'
   - CI_WEB='false'
   - CI_DELAY='120'
-  - CI_DOCKERENV=''
+  - CI_DOCKERENV='PIPER_VOICE=en_US-lessac-medium'
   - CI_AUTH=''
   - CI_WEBPATH=''

readme-vars.yml

@@ -28,11 +28,12 @@ param_ports:
 # optional container parameters
 opt_param_usage_include_env: true
 opt_param_env_vars:
+  - {env_var: "LOCAL_ONLY", env_value: "", desc: "If set to `true`, or any other value, the container will not attempt to download models from HuggingFace and will only use locally-provided models."}
   - {env_var: "PIPER_LENGTH", env_value: "1.0", desc: "Voice speaking rate, 1.0 is default with < 1.0 being faster and > 1.0 being slower."}
   - {env_var: "PIPER_NOISE", env_value: "0.667", desc: "Controls the variability of the voice by adding noise. Values above 1 will start to degrade audio."}
   - {env_var: "PIPER_NOISEW", env_value: "0.333", desc: "Controls the variability of speaking cadence. Values above 1 produce extreme stutters and pauses."}
   - {env_var: "PIPER_SPEAKER", env_value: "0", desc: "Speaker number to use if the voice supports multiple speakers."}
-  - {env_var: "PIPER_PROCS", env_value: "1", desc: "Number of Piper processes to run simultaneously."}
+  - {env_var: "STREAMING", env_value: "", desc: "Setting to `true`, or any other value, will enable support for streaming audio on sentence boundaries."}
 readonly_supported: true
 # application setup block
 app_setup_block_enabled: true
@@ -85,6 +86,7 @@ init_diagram: |
   "piper:gpu" <- Base Images
 # changelog
 changelogs:
+  - {date: "29.08.25:", desc: "Add streaming support & support for local only mode."}
   - {date: "26.11.24:", desc: "Add GPU branch."}
   - {date: "18.07.24:", desc: "Rebase to Ubuntu Noble."}
   - {date: "25.11.23:", desc: "Initial Release."}

root/etc/s6-overlay/s6-rc.d/svc-piper/run

@@ -9,19 +9,25 @@ for dir in "${NVIDIA_PACKAGE_DIR}"/*; do
     fi
 done
 
+unset UPDATE_MODELS
+
+if [[ -z ${LOCAL_ONLY} ]]; then
+    UPDATE_MODELS=true
+fi
+
 LD_LIBRARY_PATH="/lsiopy/lib/python3.12/site-packages/onnxruntime/capi:$LD_LIBRARY_PATH"
 
 exec \
     s6-notifyoncheck -d -n 300 -w 1000 -c "nc -z localhost 10200" \
         s6-setuidgid abc python3 -m wyoming_piper \
-        --piper '/usr/share/piper/piper' \
         --uri 'tcp://0.0.0.0:10200' \
         --length-scale "${PIPER_LENGTH:-1.0}" \
         --noise-scale "${PIPER_NOISE:-0.667}" \
         --noise-w "${PIPER_NOISEW:-0.333}" \
         --speaker "${PIPER_SPEAKER:-0}" \
         --voice "${PIPER_VOICE}" \
-        --max-piper-procs "${PIPER_PROCS:-1}" \
         --data-dir /config \
-        --download-dir /config --update-voices \
-        --cuda
+        --download-dir /config \
+        --use-cuda \
+        ${UPDATE_MODELS:+--update-voices} \
+        ${STREAMING:+--streaming}