Spaces:
Paused
Paused
Update tts/tts_google.py
Browse files- tts/tts_google.py +64 -64
tts/tts_google.py
CHANGED
|
@@ -1,65 +1,65 @@
|
|
| 1 |
-
# tts_google.py
|
| 2 |
-
from google.cloud import texttospeech
|
| 3 |
-
from ssml_converter import SSMLConverter
|
| 4 |
-
from utils.logger import log_info, log_error, log_debug, log_warning
|
| 5 |
-
|
| 6 |
-
class GoogleCloudTTS(TTSInterface):
|
| 7 |
-
"""Google Cloud Text-to-Speech implementation"""
|
| 8 |
-
|
| 9 |
-
def __init__(self, credentials_path: str):
|
| 10 |
-
super().__init__()
|
| 11 |
-
self.supports_ssml = True
|
| 12 |
-
self.credentials_path = credentials_path
|
| 13 |
-
|
| 14 |
-
# Google TTS doesn't need preprocessing with SSML
|
| 15 |
-
self.preprocessing_flags = set()
|
| 16 |
-
|
| 17 |
-
# Initialize client
|
| 18 |
-
os.environ['GOOGLE_APPLICATION_CREDENTIALS'] = credentials_path
|
| 19 |
-
self.client = texttospeech.TextToSpeechClient()
|
| 20 |
-
|
| 21 |
-
# SSML converter
|
| 22 |
-
self.ssml_converter = SSMLConverter(language="tr-TR")
|
| 23 |
-
|
| 24 |
-
async def synthesize(self, text: str, voice_id: Optional[str] = None, **kwargs) -> bytes:
|
| 25 |
-
"""Convert text to speech using Google Cloud TTS"""
|
| 26 |
-
try:
|
| 27 |
-
# Check if SSML should be used
|
| 28 |
-
use_ssml = kwargs.get("use_ssml", True)
|
| 29 |
-
|
| 30 |
-
if use_ssml and not text.startswith("<speak>"):
|
| 31 |
-
# Convert to SSML
|
| 32 |
-
text = self.ssml_converter.convert_to_ssml(text)
|
| 33 |
-
log_info(f"π Converted to SSML: {text[:200]}...")
|
| 34 |
-
input_text = texttospeech.SynthesisInput(ssml=text)
|
| 35 |
-
else:
|
| 36 |
-
input_text = texttospeech.SynthesisInput(text=text)
|
| 37 |
-
|
| 38 |
-
# Voice selection
|
| 39 |
-
voice = texttospeech.VoiceSelectionParams(
|
| 40 |
-
language_code=kwargs.get("language_code", "tr-TR"),
|
| 41 |
-
name=voice_id or "tr-TR-Wavenet-B",
|
| 42 |
-
ssml_gender=texttospeech.SsmlVoiceGender.FEMALE
|
| 43 |
-
)
|
| 44 |
-
|
| 45 |
-
# Audio config
|
| 46 |
-
audio_config = texttospeech.AudioConfig(
|
| 47 |
-
audio_encoding=texttospeech.AudioEncoding.MP3,
|
| 48 |
-
speaking_rate=kwargs.get("speaking_rate", 1.0),
|
| 49 |
-
pitch=kwargs.get("pitch", 0.0),
|
| 50 |
-
volume_gain_db=kwargs.get("volume_gain_db", 0.0)
|
| 51 |
-
)
|
| 52 |
-
|
| 53 |
-
# Perform synthesis
|
| 54 |
-
response = self.client.synthesize_speech(
|
| 55 |
-
input=input_text,
|
| 56 |
-
voice=voice,
|
| 57 |
-
audio_config=audio_config
|
| 58 |
-
)
|
| 59 |
-
|
| 60 |
-
log_info(f"β
Google TTS returned {len(response.audio_content)} bytes")
|
| 61 |
-
return response.audio_content
|
| 62 |
-
|
| 63 |
-
except Exception as e:
|
| 64 |
-
log_error("β Google TTS error", e)
|
| 65 |
raise
|
|
|
|
| 1 |
+
# tts_google.py
|
| 2 |
+
from google.cloud import texttospeech
|
| 3 |
+
from .ssml_converter import SSMLConverter
|
| 4 |
+
from utils.logger import log_info, log_error, log_debug, log_warning
|
| 5 |
+
|
| 6 |
+
class GoogleCloudTTS(TTSInterface):
|
| 7 |
+
"""Google Cloud Text-to-Speech implementation"""
|
| 8 |
+
|
| 9 |
+
def __init__(self, credentials_path: str):
|
| 10 |
+
super().__init__()
|
| 11 |
+
self.supports_ssml = True
|
| 12 |
+
self.credentials_path = credentials_path
|
| 13 |
+
|
| 14 |
+
# Google TTS doesn't need preprocessing with SSML
|
| 15 |
+
self.preprocessing_flags = set()
|
| 16 |
+
|
| 17 |
+
# Initialize client
|
| 18 |
+
os.environ['GOOGLE_APPLICATION_CREDENTIALS'] = credentials_path
|
| 19 |
+
self.client = texttospeech.TextToSpeechClient()
|
| 20 |
+
|
| 21 |
+
# SSML converter
|
| 22 |
+
self.ssml_converter = SSMLConverter(language="tr-TR")
|
| 23 |
+
|
| 24 |
+
async def synthesize(self, text: str, voice_id: Optional[str] = None, **kwargs) -> bytes:
|
| 25 |
+
"""Convert text to speech using Google Cloud TTS"""
|
| 26 |
+
try:
|
| 27 |
+
# Check if SSML should be used
|
| 28 |
+
use_ssml = kwargs.get("use_ssml", True)
|
| 29 |
+
|
| 30 |
+
if use_ssml and not text.startswith("<speak>"):
|
| 31 |
+
# Convert to SSML
|
| 32 |
+
text = self.ssml_converter.convert_to_ssml(text)
|
| 33 |
+
log_info(f"π Converted to SSML: {text[:200]}...")
|
| 34 |
+
input_text = texttospeech.SynthesisInput(ssml=text)
|
| 35 |
+
else:
|
| 36 |
+
input_text = texttospeech.SynthesisInput(text=text)
|
| 37 |
+
|
| 38 |
+
# Voice selection
|
| 39 |
+
voice = texttospeech.VoiceSelectionParams(
|
| 40 |
+
language_code=kwargs.get("language_code", "tr-TR"),
|
| 41 |
+
name=voice_id or "tr-TR-Wavenet-B",
|
| 42 |
+
ssml_gender=texttospeech.SsmlVoiceGender.FEMALE
|
| 43 |
+
)
|
| 44 |
+
|
| 45 |
+
# Audio config
|
| 46 |
+
audio_config = texttospeech.AudioConfig(
|
| 47 |
+
audio_encoding=texttospeech.AudioEncoding.MP3,
|
| 48 |
+
speaking_rate=kwargs.get("speaking_rate", 1.0),
|
| 49 |
+
pitch=kwargs.get("pitch", 0.0),
|
| 50 |
+
volume_gain_db=kwargs.get("volume_gain_db", 0.0)
|
| 51 |
+
)
|
| 52 |
+
|
| 53 |
+
# Perform synthesis
|
| 54 |
+
response = self.client.synthesize_speech(
|
| 55 |
+
input=input_text,
|
| 56 |
+
voice=voice,
|
| 57 |
+
audio_config=audio_config
|
| 58 |
+
)
|
| 59 |
+
|
| 60 |
+
log_info(f"β
Google TTS returned {len(response.audio_content)} bytes")
|
| 61 |
+
return response.audio_content
|
| 62 |
+
|
| 63 |
+
except Exception as e:
|
| 64 |
+
log_error("β Google TTS error", e)
|
| 65 |
raise
|