feat: first commit

.dockerignore

@@ -0,0 +1,3 @@
+.venv
+.git
+data

.gitignore

@@ -0,0 +1,117 @@
+output/*
+
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+pip-wheel-metadata/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+.hypothesis/
+.pytest_cache/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don’t work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# celery beat schedule file
+celerybeat-schedule
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# Pycharm
+.idea
+
+data

.pre-commit-config.yaml

@@ -0,0 +1,27 @@
+---
+repos:
+  - repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v2.3.0
+    hooks:
+      - id: check-yaml
+        args: [--allow-multiple-documents]
+      - id: end-of-file-fixer
+      - id: trailing-whitespace
+  - repo: local
+    hooks:
+      - id: black
+        name: black
+        entry: poetry run black src
+        language: system
+        types: [python]
+        pass_filenames: false
+      - id: ruff
+        name: ruff
+        entry: poetry run ruff --fix src
+        language: system
+        types: [python]
+        pass_filenames: false
+  - repo: https://github.com/pre-commit/mirrors-prettier
+    rev: v2.3.2
+    hooks:
+      - id: prettier

Dockerfile

@@ -0,0 +1,9 @@
+FROM rocm/pytorch:rocm5.7_ubuntu22.04_py3.10_pytorch_2.0.1
+
+RUN pip install -U pip && pip install wyoming==1.2.0 openai-whisper==20231117 tokenizers==0.13.*
+
+COPY src /src
+
+WORKDIR /src
+
+ENTRYPOINT ["/src/run.sh"]

README.md

@@ -0,0 +1,29 @@
+# rocm-wyoming-whisper
+
+A docker image and a few lines of python to use OpenAI whisper with Rhasspy and/or Home Assistant on AMD GPUs with ROCm.
+
+## Run with docker-compose
+
+```shell
+docker-compose up -d
+```
+
+## Run with Docker
+
+Build docker image:
+
+```shell
+docker build -t wyoming-whisper .
+```
+
+Run docker image:
+
+```shell
+docker run --entrypoint '' -v $(pwd)/data:/data -v $(pwd)/src:/src -it --device=/dev/kfd --device=/dev/dri --security-opt seccomp=unconfined --group-add video -p 10300:10300 wyoming-whisper bash
+```
+
+Run script:
+
+```shell
+python -m wyoming_whisper --download-dir /data --model medium --debug
+```

docker-compose.yml

@@ -0,0 +1,22 @@
+services:
+  whisper:
+    container_name: whisper
+    restart: unless-stopped
+    build:
+      context: .
+    group_add:
+      - video
+    volumes:
+      - ./data:/data
+    command:
+      - --download-dir
+      - /data
+      - --language
+      - en
+      - --device
+      - cuda
+    ports:
+      - 10300:10300
+    devices:
+      - "/dev/dri"
+      - "/dev/kfd"

pyproject.toml

@@ -0,0 +1,18 @@
+[tool.poetry]
+name = "wyoming-whisper"
+version = "0.1.0"
+description = ""
+authors = ["Simon Guigui <simon.guigui@parkoview.com>"]
+readme = "README.md"
+
+[tool.poetry.dependencies]
+python = "^3.10"
+openai-whisper = "20231117"
+wyoming = "^1.2.0"
+tokenizers = "0.13.*"
+black = "^23.11.0"
+ruff = "^0.1.6"
+
+[build-system]
+requires = ["poetry-core"]
+build-backend = "poetry.core.masonry.api"

src/run.sh

@@ -0,0 +1,3 @@
+#!/usr/bin/env bash
+
+exec python3 -m wyoming_whisper --download-dir /data "$@"

src/wyoming_whisper/__init__.py

@@ -0,0 +1 @@
+"""Wyoming server for faster-whisper."""

src/wyoming_whisper/__main__.py

@@ -0,0 +1,103 @@
+import argparse
+import asyncio
+import logging
+from functools import partial
+
+from wyoming.info import AsrModel, AsrProgram, Attribution, Info
+from wyoming.server import AsyncServer
+
+from .const import WHISPER_LANGUAGES
+import whisper
+from .handler import FasterWhisperEventHandler
+
+_LOGGER = logging.getLogger(__name__)
+
+
+async def main() -> None:
+    """Main entry point."""
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--model",
+        default="medium",
+        help="Name of whisper model to use (default medium)",
+    )
+    parser.add_argument(
+        "--uri", help="unix:// or tcp://", default="tcp://0.0.0.0:10300"
+    )
+    parser.add_argument(
+        "--download-dir",
+        required=True,
+        help="Directory to download models into",
+    )
+    parser.add_argument(
+        "--device",
+        default="cuda",
+        help="Device to use for inference (default: cuda)",
+    )
+    parser.add_argument(
+        "--language",
+        default="en",
+        help="Default language to set for transcription (default: en)",
+    )
+    parser.add_argument(
+        "--beam-size",
+        type=int,
+        default=5,
+    )
+
+    parser.add_argument("--debug", action="store_true", help="Log DEBUG messages")
+    args = parser.parse_args()
+
+    logging.basicConfig(level=logging.DEBUG if args.debug else logging.INFO)
+    _LOGGER.debug(args)
+
+    _LOGGER.debug("Loading %s", args.model)
+    whisper_model = whisper.load_model(
+        args.model, device=args.device, download_root=args.download_dir
+    )
+
+    wyoming_info = Info(
+        asr=[
+            AsrProgram(
+                name="whisper",
+                description="Whisper",
+                attribution=Attribution(
+                    name="OpenAI",
+                    url="https://github.com/openai/whisper",
+                ),
+                installed=True,
+                models=[
+                    AsrModel(
+                        name=args.model,
+                        description=args.model,
+                        attribution=Attribution(
+                            name="whisper",
+                            url="https://github.com/openai/whisper",
+                        ),
+                        installed=True,
+                        languages=WHISPER_LANGUAGES,
+                    )
+                ],
+            )
+        ],
+    )
+
+    server = AsyncServer.from_uri(args.uri)
+    _LOGGER.info("Ready")
+    model_lock = asyncio.Lock()
+    await server.run(
+        partial(
+            FasterWhisperEventHandler,
+            wyoming_info,
+            args,
+            whisper_model,
+            model_lock,
+        )
+    )
+
+
+if __name__ == "__main__":
+    try:
+        asyncio.run(main())
+    except KeyboardInterrupt:
+        pass

src/wyoming_whisper/const.py

@@ -0,0 +1,101 @@
+WHISPER_LANGUAGES = [
+    "af",
+    "am",
+    "ar",
+    "as",
+    "az",
+    "ba",
+    "be",
+    "bg",
+    "bn",
+    "bo",
+    "br",
+    "bs",
+    "ca",
+    "cs",
+    "cy",
+    "da",
+    "de",
+    "el",
+    "en",
+    "es",
+    "et",
+    "eu",
+    "fa",
+    "fi",
+    "fo",
+    "fr",
+    "gl",
+    "gu",
+    "ha",
+    "haw",
+    "he",
+    "hi",
+    "hr",
+    "ht",
+    "hu",
+    "hy",
+    "id",
+    "is",
+    "it",
+    "ja",
+    "jw",
+    "ka",
+    "kk",
+    "km",
+    "kn",
+    "ko",
+    "la",
+    "lb",
+    "ln",
+    "lo",
+    "lt",
+    "lv",
+    "mg",
+    "mi",
+    "mk",
+    "ml",
+    "mn",
+    "mr",
+    "ms",
+    "mt",
+    "my",
+    "ne",
+    "nl",
+    "nn",
+    "no",
+    "oc",
+    "pa",
+    "pl",
+    "ps",
+    "pt",
+    "ro",
+    "ru",
+    "sa",
+    "sd",
+    "si",
+    "sk",
+    "sl",
+    "sn",
+    "so",
+    "sq",
+    "sr",
+    "su",
+    "sv",
+    "sw",
+    "ta",
+    "te",
+    "tg",
+    "th",
+    "tk",
+    "tl",
+    "tr",
+    "tt",
+    "uk",
+    "ur",
+    "uz",
+    "vi",
+    "yi",
+    "yo",
+    "zh",
+]

src/wyoming_whisper/handler.py

@@ -0,0 +1,92 @@
+"""Event handler for clients of the server."""
+import argparse
+import asyncio
+import logging
+
+import numpy as np
+from wyoming.asr import Transcribe, Transcript
+from wyoming.audio import AudioChunk, AudioChunkConverter, AudioStop
+from wyoming.event import Event
+from wyoming.info import Describe, Info
+from wyoming.server import AsyncEventHandler
+
+from whisper import Whisper
+
+_LOGGER = logging.getLogger(__name__)
+
+
+class FasterWhisperEventHandler(AsyncEventHandler):
+    """Event handler for clients."""
+
+    def __init__(
+        self,
+        wyoming_info: Info,
+        cli_args: argparse.Namespace,
+        model: Whisper,
+        model_lock: asyncio.Lock,
+        *args,
+        **kwargs,
+    ) -> None:
+        super().__init__(*args, **kwargs)
+
+        self.cli_args = cli_args
+        self.wyoming_info_event = wyoming_info.event()
+        self.model = model
+        self.model_lock = model_lock
+        self.audio = bytes()
+        self.audio_converter = AudioChunkConverter(
+            rate=16000,
+            width=2,
+            channels=1,
+        )
+        self._language = self.cli_args.language
+
+    async def handle_event(self, event: Event) -> bool:
+        if Describe.is_type(event.type):
+            await self.write_event(self.wyoming_info_event)
+            _LOGGER.debug("Sent info")
+            return True
+
+        if Transcribe.is_type(event.type):
+            transcribe = Transcribe.from_event(event)
+            if transcribe.language:
+                self._language = transcribe.language
+                _LOGGER.debug("Language set to %s", transcribe.language)
+            return True
+
+        if AudioChunk.is_type(event.type):
+            if not self.audio:
+                _LOGGER.debug("Receiving audio")
+
+            chunk = AudioChunk.from_event(event)
+            chunk = self.audio_converter.convert(chunk)
+            self.audio += chunk.audio
+
+            return True
+
+        if AudioStop.is_type(event.type):
+            _LOGGER.debug("Audio stopped")
+            audio = (
+                np.frombuffer(self.audio, dtype=np.int16).astype(np.float32) / 32678.0
+            )
+            async with self.model_lock:
+                response = self.model.transcribe(
+                    audio,
+                    beam_size=self.cli_args.beam_size,
+                    language=self._language,
+                )
+            segments = response["segments"]
+
+            text = " ".join(segment["text"] for segment in segments)
+            _LOGGER.info(text)
+
+            await self.write_event(Transcript(text=text).event())
+            _LOGGER.debug("Completed request")
+
+            # Reset
+            self.audio = bytes()
+            self._language = self.cli_args.language
+
+            return False
+
+        return True