feat(hermes): Piper TTS (local US male, no cloud) #17
@@ -52,7 +52,7 @@ 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 ----------
|
||||
# ---------- Patch tts_tool.py: replace Edge TTS fallback with Piper ----------
|
||||
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
|
||||
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
#!/usr/bin/env python3
|
||||
"""Patch Hermes TTS tool to add Piper provider and remove Edge TTS fallback."""
|
||||
"""Patch Hermes TTS tool: remove Edge TTS, replace with Piper as default/fallback."""
|
||||
import sys
|
||||
|
||||
tts_path = '/opt/hermes/.venv/lib/python3.13/site-packages/tools/tts_tool.py'
|
||||
@@ -7,58 +7,6 @@ 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
|
||||
@@ -89,8 +37,8 @@ old_edge = ''' else:
|
||||
"or set up NeuTTS for local synthesis."
|
||||
}, ensure_ascii=False)'''
|
||||
|
||||
new_piper_fallback = ''' else:
|
||||
# Default: Piper TTS (local, CPU, no cloud)
|
||||
new_piper = ''' else:
|
||||
# Default: Piper TTS (local, CPU, no cloud, no Microsoft)
|
||||
piper_available = False
|
||||
try:
|
||||
piper_binary = "/opt/hermes/.venv/bin/piper"
|
||||
@@ -127,19 +75,20 @@ new_piper_fallback = ''' else:
|
||||
}, ensure_ascii=False)'''
|
||||
|
||||
if old_edge in code:
|
||||
code = code.replace(old_edge, new_piper_fallback)
|
||||
code = code.replace(old_edge, new_piper)
|
||||
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")
|
||||
print("Piper fallback already present")
|
||||
else:
|
||||
print("ERROR: Could not find Edge fallback")
|
||||
# Debug: show what the else block looks like
|
||||
print("ERROR: Could not find Edge fallback in tts_tool.py")
|
||||
# Debug output
|
||||
import re
|
||||
match = re.search(r' else:\n # Default:', code)
|
||||
if match:
|
||||
print("Found else block at", match.start())
|
||||
for m in re.finditer(r' else:\n # Default:', code):
|
||||
start = max(0, m.start() - 100)
|
||||
end = min(len(code), m.end() + 200)
|
||||
print(f"Found else/default at position {m.start()}:")
|
||||
print(code[start:end])
|
||||
sys.exit(1)
|
||||
|
||||
with open(tts_path, 'w') as f:
|
||||
|
||||
Reference in New Issue
Block a user