test initilaize

Dockerfile

@@ -1,13 +1,53 @@
-FROM python:3.12
+FROM python:3.11-slim
 
-RUN useradd -m -u 1000 user
-USER user
-ENV PATH="/home/user/.local/bin:$PATH"
+ENV PYTHONDONTWRITEBYTECODE=1 \
+    PYTHONUNBUFFERED=1 \
+    PIP_NO_CACHE_DIR=1 \
+    PATH="/home/user/.local/bin:$PATH" \
+    HF_HOME="/home/user/.cache/huggingface" \
+    WHISPER_CACHE_DIR="/home/user/.cache/whisper"
 
-WORKDIR /app
+RUN apt-get update \
+    && apt-get install -y --no-install-recommends \
+        build-essential \
+        ffmpeg \
+        git \
+        libsndfile1 \
+    && rm -rf /var/lib/apt/lists/*
+
+ARG WHISPERLIVEKIT_REPO="https://github.com/Alimjoo/WhisperLiveKit-ug.git"
+ARG WHISPERLIVEKIT_REF=""
+
+RUN git clone --depth 1 "${WHISPERLIVEKIT_REPO}" /opt/WhisperLiveKit-ug \
+    && if [ -n "${WHISPERLIVEKIT_REF}" ]; then \
+        cd /opt/WhisperLiveKit-ug \
+        && git fetch --depth 1 origin "${WHISPERLIVEKIT_REF}" \
+        && git checkout FETCH_HEAD; \
+    fi
+
+WORKDIR /opt/WhisperLiveKit-ug
+RUN pip install --upgrade pip setuptools wheel \
+    && pip install -e .
+
+RUN useradd -m -u 1000 user \
+    && mkdir -p /app /home/user/models "${HF_HOME}" "${WHISPER_CACHE_DIR}" \
+    && chown -R user:user /app /home/user /opt/WhisperLiveKit-ug
 
+WORKDIR /app
 COPY --chown=user ./requirements.txt requirements.txt
-RUN pip install --no-cache-dir --upgrade -r requirements.txt
+RUN pip install -r requirements.txt
 
 COPY --chown=user . /app
-CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]
+RUN chmod +x /app/start.sh
+
+USER user
+
+ENV HF_MODEL_REPO="piyazon/whisper_uyghur_pt" \
+    HF_MODEL_FILENAMES="uyghur_whisper_base.pt uyghur_whisper_tiny.pt" \
+    WHISPER_MODEL_FILENAME="uyghur_whisper_base.pt" \
+    WHISPER_LANGUAGE="ug" \
+    WHISPER_BACKEND="whisper" \
+    PORT="7860"
+
+EXPOSE 7860
+CMD ["/app/start.sh"]

README.md

@@ -4,7 +4,32 @@ emoji: 🔥
 colorFrom: pink
 colorTo: gray
 sdk: docker
+app_port: 7860
 pinned: false
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# ASR Streaming Server
+
+Docker Space for running `https://github.com/Alimjoo/WhisperLiveKit-ug.git`
+with Uyghur Whisper `.pt` models from `piyazon/whisper_uyghur_pt`.
+
+## Space secrets
+
+Set `HF_TOKEN` as a Space secret with read access to the private model repo.
+`HUGGING_FACE_HUB_TOKEN` is also supported.
+
+## Defaults
+
+- WhisperLiveKit repo: `https://github.com/Alimjoo/WhisperLiveKit-ug.git`
+- Hugging Face model repo: `piyazon/whisper_uyghur_pt`
+- Downloaded model files: `uyghur_whisper_base.pt uyghur_whisper_tiny.pt`
+- Started model: `uyghur_whisper_base.pt`
+- Port: `7860`
+
+## Runtime configuration
+
+- `WHISPER_MODEL_FILENAME=uyghur_whisper_tiny.pt` starts the tiny model instead.
+- `WHISPER_MODEL_PATH=/path/to/model.pt` skips Hugging Face download.
+- `WHISPER_LANGUAGE=auto` can be used if the fork rejects `ug`.
+- `WHISPERLIVEKIT_ARGS` appends extra CLI arguments, for example
+  `--backend-policy localagreement`.

download_models.py

@@ -0,0 +1,62 @@
+import os
+import sys
+from pathlib import Path
+
+from huggingface_hub import hf_hub_download
+
+
+DEFAULT_MODEL_REPO = "piyazon/whisper_uyghur_pt"
+DEFAULT_MODEL_FILENAMES = "uyghur_whisper_base.pt uyghur_whisper_tiny.pt"
+DEFAULT_SELECTED_MODEL = "uyghur_whisper_base.pt"
+
+
+def _split_filenames(value: str) -> list[str]:
+    return [item.strip() for item in value.replace(",", " ").split() if item.strip()]
+
+
+def _log(message: str) -> None:
+    print(message, file=sys.stderr, flush=True)
+
+
+def main() -> int:
+    repo_id = os.getenv("HF_MODEL_REPO", DEFAULT_MODEL_REPO)
+    selected_filename = os.getenv("WHISPER_MODEL_FILENAME", DEFAULT_SELECTED_MODEL)
+    model_dir = Path(os.getenv("MODEL_DIR", "/home/user/models")).expanduser()
+    token = os.getenv("HF_TOKEN") or os.getenv("HUGGING_FACE_HUB_TOKEN")
+
+    filenames = _split_filenames(os.getenv("HF_MODEL_FILENAMES", DEFAULT_MODEL_FILENAMES))
+    if selected_filename not in filenames:
+        filenames.insert(0, selected_filename)
+
+    model_dir.mkdir(parents=True, exist_ok=True)
+
+    downloaded_paths: dict[str, str] = {}
+    for filename in filenames:
+        cached_path = model_dir / filename
+        if cached_path.exists() and cached_path.stat().st_size > 0:
+            _log(f"Using cached model: {cached_path}")
+            downloaded_paths[filename] = str(cached_path)
+            continue
+
+        _log(f"Downloading {repo_id}/{filename} to {model_dir}")
+        try:
+            downloaded_paths[filename] = hf_hub_download(
+                repo_id=repo_id,
+                filename=filename,
+                local_dir=str(model_dir),
+                token=token,
+            )
+        except Exception as exc:
+            _log(f"Failed to download {repo_id}/{filename}: {exc}")
+            _log(
+                "If the model repository is private, add an HF_TOKEN Space secret "
+                "with read access to the model repo."
+            )
+            return 1
+
+    print(downloaded_paths[selected_filename], flush=True)
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

requirements.txt

@@ -1,2 +1,4 @@
 fastapi
-uvicorn[standard]
+uvicorn[standard]
+huggingface_hub>=0.23
+hf_xet>=1.0

start.sh

@@ -0,0 +1,48 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+PORT="${PORT:-7860}"
+MODEL_PATH="${WHISPER_MODEL_PATH:-}"
+WHISPER_LANGUAGE="${WHISPER_LANGUAGE:-ug}"
+WHISPER_BACKEND="${WHISPER_BACKEND:-whisper}"
+WHISPERLIVEKIT_ARGS="${WHISPERLIVEKIT_ARGS:-}"
+
+if [ -z "${MODEL_PATH}" ]; then
+    MODEL_PATH="$(python /app/download_models.py)"
+fi
+
+if command -v wlk >/dev/null 2>&1; then
+    WLK_CMD="wlk"
+elif command -v whisperlivekit-server >/dev/null 2>&1; then
+    WLK_CMD="whisperlivekit-server"
+else
+    echo "Neither 'wlk' nor 'whisperlivekit-server' is available on PATH." >&2
+    exit 1
+fi
+
+HELP_FILE="/tmp/whisperlivekit-help.txt"
+"${WLK_CMD}" --help >"${HELP_FILE}" 2>&1 || true
+
+ARGS=(--host 0.0.0.0 --port "${PORT}")
+
+if grep -q -- "--model-path" "${HELP_FILE}"; then
+    ARGS+=(--model-path "${MODEL_PATH}")
+else
+    ARGS+=(--model "${MODEL_PATH}")
+fi
+
+if [ -n "${WHISPER_LANGUAGE}" ] && grep -q -- "--language" "${HELP_FILE}"; then
+    ARGS+=(--language "${WHISPER_LANGUAGE}")
+fi
+
+if [ -n "${WHISPER_BACKEND}" ] && grep -q -- "--backend" "${HELP_FILE}"; then
+    ARGS+=(--backend "${WHISPER_BACKEND}")
+fi
+
+if [ -n "${WHISPERLIVEKIT_ARGS}" ]; then
+    read -r -a EXTRA_ARGS <<<"${WHISPERLIVEKIT_ARGS}"
+    ARGS+=("${EXTRA_ARGS[@]}")
+fi
+
+echo "Starting WhisperLiveKit on 0.0.0.0:${PORT} with model ${MODEL_PATH}"
+exec "${WLK_CMD}" "${ARGS[@]}"