diff --git a/ai/Dockerfile b/ai/Dockerfile index 7de3171..87bdd34 100644 --- a/ai/Dockerfile +++ b/ai/Dockerfile @@ -38,49 +38,24 @@ USER hermes # On copie tout le projet d'un coup sans assumer la présence de fichiers de lock spécifiques COPY --chown=hermes:hermes . . -# ---------- Python virtualenv ---------- +# ---------- Python virtualenv avec Piper TTS ---------- RUN uv venv && \ - uv pip install --no-cache-dir sounddevice numpy faster-whisper + uv pip install --no-cache-dir piper-tts sounddevice numpy faster-whisper -# ---------- Patch tts_tool.py to add Coqui provider ---------- -RUN /opt/hermes/.venv/bin/python3 -c " -tts_path = '/opt/hermes/.venv/lib/python3.13/site-packages/tools/tts_tool.py' -with open(tts_path) as f: - code = f.read() -coqui_block = ''' - elif provider == \"coqui\": - logger.info(\"Generating speech with Coqui TTS (GPU, local)...\") - import subprocess - coqui_python = \"/opt/coqui-tts/bin/python3\" - coqui_script = \"/opt/coqui-tts/bin/coqui_synth.py\" - coqui_config = tts_config.get(\"coqui\", {}) - model = coqui_config.get(\"model\", \"tts_models/en/vctk/vits\") - use_gpu = coqui_config.get(\"use_gpu\", True) - speaker = coqui_config.get(\"speaker\", \"\") - cmd = [ - coqui_python, coqui_script, - \"--text\", text, - \"--out\", file_str, - \"--model\", model, - ] - if use_gpu: - cmd.append(\"--gpu\") - if speaker: - cmd.extend([\"--speaker\", speaker]) - result = subprocess.run(cmd, capture_output=True, text=True, timeout=120) - if result.returncode != 0: - stderr = result.stderr.strip() - raise RuntimeError(f\"Coqui TTS failed: {stderr or 'unknown error'}\") - logger.info(\"Coqui TTS audio saved: %s\", file_str) -''' -code = code.replace( - ' else:\n # Default: Edge TTS (free), with NeuTTS as local fallback', - coqui_block + ' else:\n # Default: Edge TTS (free), with NeuTTS as local fallback' -) -with open(tts_path, 'w') as f: - f.write(code) +# ---------- Télécharger la voix Piper Ryan ---------- +RUN mkdir -p /opt/hermes/.venv/share/piper/voices && \ + /opt/hermes/.venv/bin/python3 -c " +import urllib.request +base = '/opt/hermes/.venv/share/piper/voices' +url = 'https://huggingface.co/rhasspy/piper-voices/resolve/main/en/en_US/ryan/high/en_US-ryan-high.onnx' +urllib.request.urlretrieve(url + '?download=true', base + '/en_US-ryan-high.onnx') +urllib.request.urlretrieve(url + '.onnx.json?download=true', base + '/en_US-ryan-high.onnx.json') " +# ---------- Patch tts_tool.py: remplacer Coqui par Piper, supprimer Edge ---------- +COPY ai/patch_tts_tool.py /tmp/patch_tts_tool.py +RUN /opt/hermes/.venv/bin/python3 /tmp/patch_tts_tool.py && rm /tmp/patch_tts_tool.py + # ---------- Runtime ---------- ENV HERMES_HOME=/opt/data ENV PATH="/opt/data/.local/bin:${PATH}" diff --git a/ai/__pycache__/patch_tts_tool.cpython-313.pyc b/ai/__pycache__/patch_tts_tool.cpython-313.pyc new file mode 100644 index 0000000..ac2d1ae Binary files /dev/null and b/ai/__pycache__/patch_tts_tool.cpython-313.pyc differ diff --git a/ai/patch_tts_tool.py b/ai/patch_tts_tool.py new file mode 100644 index 0000000..4791132 --- /dev/null +++ b/ai/patch_tts_tool.py @@ -0,0 +1,147 @@ +#!/usr/bin/env python3 +"""Patch Hermes TTS tool to add Piper provider and remove Edge TTS fallback.""" +import sys + +tts_path = '/opt/hermes/.venv/lib/python3.13/site-packages/tools/tts_tool.py' + +with open(tts_path) as f: + code = f.read() + +# Replace the Coqui provider block with Piper +old_coqui = ' elif provider == "coqui":' +new_piper = ''' elif provider == "piper": + logger.info("Generating speech with Piper TTS (local, CPU)...") + import subprocess + piper_binary = "/opt/hermes/.venv/bin/piper" + piper_config = tts_config.get("piper", {}) + voice = piper_config.get("voice", "en_US-lessac-medium") + model_dir = piper_config.get("model_dir", "/opt/hermes/.venv/share/piper/voices") + model_path = os.path.join(model_dir, f"{voice}.onnx") + if not os.path.exists(model_path): + raise FileNotFoundError(f"Piper voice model not found: {model_path}") + cmd = [piper_binary, "--model", model_path, "--output-raw"] + proc = subprocess.Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE) + raw_audio, stderr = proc.communicate(input=text.encode(), timeout=60) + if proc.returncode != 0: + raise RuntimeError(f"Piper TTS failed: {stderr.decode()[:200]}") + ffmpeg_cmd = ["ffmpeg", "-f", "s16le", "-ar", "22050", "-ac", "1", "-i", "-", "-y", file_str] + subprocess.run(ffmpeg_cmd, input=raw_audio, capture_output=True, timeout=30) + logger.info("Piper TTS audio saved: %s", file_str)''' + +if old_coqui in code: + code = code.replace(old_coqui, new_piper) + print("Coqui -> Piper replaced") +else: + # Fresh Hermes install - add Piper after the last provider (kittentts) + if 'provider == "piper"' in code: + print("Piper already present, skipping coqui replacement") + else: + print("Stock Hermes - adding Piper provider after kittentts...") + marker = ' elif provider == "kittentts":' + if marker in code: + # Find the end of the kittentts block and insert Piper before else + lines = code.split('\n') + kit_idx = None + for i, line in enumerate(lines): + if line.strip().startswith('elif provider == "kittentts":'): + kit_idx = i + break + if kit_idx is not None: + # Find the next blank line + else block + for i in range(kit_idx, len(lines)): + if lines[i].strip() == 'else:': + # Insert Piper block before this line + indent = ' ' + piper_lines = new_piper.split('\n') + insert = piper_lines + [''] + lines[i:i] = insert + code = '\n'.join(lines) + print("Piper provider added after kittentts") + break + +# Replace the Edge fallback with Piper fallback +old_edge = ''' else: + # Default: Edge TTS (free), with NeuTTS as local fallback + edge_available = True + try: + _import_edge_tts() + except ImportError: + edge_available = False + + if edge_available: + logger.info("Generating speech with Edge TTS...") + try: + import concurrent.futures + with concurrent.futures.ThreadPoolExecutor(max_workers=1) as pool: + pool.submit( + lambda: asyncio.run(_generate_edge_tts(text, file_str, tts_config)) + ).result(timeout=60) + except RuntimeError: + asyncio.run(_generate_edge_tts(text, file_str, tts_config)) + elif _check_neutts_available(): + logger.info("Edge TTS not available, falling back to NeuTTS (local)...") + provider = "neutts" + _generate_neutts(text, file_str, tts_config) + else: + return json.dumps({ + "success": False, + "error": "No TTS provider available. Install edge-tts (pip install edge-tts) " + "or set up NeuTTS for local synthesis." + }, ensure_ascii=False)''' + +new_piper_fallback = ''' else: + # Default: Piper TTS (local, CPU, no cloud) + piper_available = False + try: + piper_binary = "/opt/hermes/.venv/bin/piper" + piper_config = tts_config.get("piper", {}) + voice = piper_config.get("voice", "en_US-lessac-medium") + model_dir = piper_config.get("model_dir", "/opt/hermes/.venv/share/piper/voices") + model_path = os.path.join(model_dir, f"{voice}.onnx") + if os.path.exists(model_path): + piper_available = True + except Exception: + pass + + if piper_available: + logger.info("Generating speech with Piper TTS (local, CPU)...") + import subprocess + piper_binary = "/opt/hermes/.venv/bin/piper" + piper_config = tts_config.get("piper", {}) + voice = piper_config.get("voice", "en_US-lessac-medium") + model_dir = piper_config.get("model_dir", "/opt/hermes/.venv/share/piper/voices") + model_path = os.path.join(model_dir, f"{voice}.onnx") + cmd = [piper_binary, "--model", model_path, "--output-raw"] + proc = subprocess.Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE) + raw_audio, stderr = proc.communicate(input=text.encode(), timeout=60) + if proc.returncode != 0: + raise RuntimeError(f"Piper TTS failed: {stderr.decode()[:200]}") + ffmpeg_cmd = ["ffmpeg", "-f", "s16le", "-ar", "22050", "-ac", "1", "-i", "-", "-y", file_str] + subprocess.run(ffmpeg_cmd, input=raw_audio, capture_output=True, timeout=30) + logger.info("Piper TTS audio saved: %s", file_str) + else: + return json.dumps({ + "success": False, + "error": "No TTS provider available. Install Piper TTS (pip install piper-tts) " + "and download a voice model." + }, ensure_ascii=False)''' + +if old_edge in code: + code = code.replace(old_edge, new_piper_fallback) + print("Edge fallback replaced with Piper") +else: + print("Edge fallback NOT found, checking if already Piper...") + if 'Default: Piper TTS' in code: + print("Piper fallback already present, skipping") + else: + print("ERROR: Could not find Edge fallback") + # Debug: show what the else block looks like + import re + match = re.search(r' else:\n # Default:', code) + if match: + print("Found else block at", match.start()) + sys.exit(1) + +with open(tts_path, 'w') as f: + f.write(code) +print("tts_tool.py patched successfully")