wx44wx/pipecat
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

inworld: largely adjustments for docstring compatibility

Browse Source
This commit is contained in:
padillamt
2025-07-21 12:07:58 -07:00
parent aadd088b50
commit 54ff946976
1 changed files with 12 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
src/pipecat/services/inworld/tts.py
View File

@@ -10,6 +10,7 @@ This module provides integration with Inworld AI's HTTP-based TTS API, enabling
real-time text-to-speech synthesis with high-quality, natural-sounding voices. real-time text-to-speech synthesis with high-quality, natural-sounding voices.
Key Features: Key Features:
- HTTP streaming API support for low-latency audio generation - HTTP streaming API support for low-latency audio generation
- Multiple voice options (Ashley, Hades, etc.) - Multiple voice options (Ashley, Hades, etc.)
- Real-time audio chunk processing with proper buffering - Real-time audio chunk processing with proper buffering
@@ -17,13 +18,14 @@ Key Features:
- Comprehensive error handling and metrics tracking - Comprehensive error handling and metrics tracking
Technical Implementation: Technical Implementation:
- Uses aiohttp for HTTP streaming connections - Uses aiohttp for HTTP streaming connections
- Implements JSON line-by-line parsing for streaming responses - Implements JSON line-by-line parsing for streaming responses
- Handles base64-encoded audio data with proper decoding - Handles base64-encoded audio data with proper decoding
- Manages audio continuity to prevent clicks and artifacts - Manages audio continuity to prevent clicks and artifacts
- Integrates with Pipecat's frame-based pipeline architecture - Integrates with Pipecat's frame-based pipeline architecture
Usage: Usage::
tts = InworldHttpTTSService( tts = InworldHttpTTSService(
api_key=os.getenv("INWORLD_API_KEY"), api_key=os.getenv("INWORLD_API_KEY"),
voice_id="Ashley", voice_id="Ashley",
@@ -71,6 +73,7 @@ def language_to_inworld_language(language: Language) -> Optional[str]:
corresponding language codes expected by Inworld's API. corresponding language codes expected by Inworld's API.
Supported Languages: Supported Languages:
- EN (English) -> "en" - EN (English) -> "en"
- ES (Spanish) -> "es" - ES (Spanish) -> "es"
- FR (French) -> "fr" - FR (French) -> "fr"
@@ -126,6 +129,7 @@ class InworldHttpTTSService(TTSService):
and low-latency streaming audio delivery. and low-latency streaming audio delivery.
Key Features: Key Features:
- Real-time HTTP streaming for minimal latency - Real-time HTTP streaming for minimal latency
- Multiple voice options (Ashley, Hades, etc.) - Multiple voice options (Ashley, Hades, etc.)
- High-quality audio output (48kHz LINEAR16 PCM) - High-quality audio output (48kHz LINEAR16 PCM)
@@ -134,6 +138,7 @@ class InworldHttpTTSService(TTSService):
- Built-in performance metrics and monitoring - Built-in performance metrics and monitoring
Technical Architecture: Technical Architecture:
- Uses aiohttp for non-blocking HTTP requests - Uses aiohttp for non-blocking HTTP requests
- Implements JSON line-by-line streaming protocol - Implements JSON line-by-line streaming protocol
- Processes base64-encoded audio chunks in real-time - Processes base64-encoded audio chunks in real-time
@@ -141,16 +146,17 @@ class InworldHttpTTSService(TTSService):
- Integrates with Pipecat's frame-based pipeline system - Integrates with Pipecat's frame-based pipeline system
Supported Configuration: Supported Configuration:
- Voice Selection: Ashley, Hades, and other Inworld voices - Voice Selection: Ashley, Hades, and other Inworld voices
- Models: inworld-tts-1 and other available models - Models: inworld-tts-1 and other available models
- Audio Formats: LINEAR16 PCM at various sample rates - Audio Formats: LINEAR16 PCM at various sample rates
- Languages: English, Spanish, French, Korean, Dutch, Chinese - Languages: English, Spanish, French, Korean, Dutch, Chinese
Example Usage: Example Usage::
async with aiohttp.ClientSession() as session: async with aiohttp.ClientSession() as session:
tts = InworldHttpTTSService( tts = InworldHttpTTSService(
api_key=os.getenv("INWORLD_API_KEY"), api_key=os.getenv("INWORLD_API_KEY"),
voice_id="Ashley", # Voice selection voice_id="Ashley", # Voice selection
model="inworld-tts-1", # TTS model model="inworld-tts-1", # TTS model
aiohttp_session=session, # Required HTTP session aiohttp_session=session, # Required HTTP session
sample_rate=48000, # Audio quality sample_rate=48000, # Audio quality
@@ -162,16 +168,9 @@ class InworldHttpTTSService(TTSService):
Parameters: Parameters:
language: Language to use for synthesis. language: Language to use for synthesis.
speed: Voice speed control (string or float).
emotion: List of emotion controls.
.. deprecated:: 0.0.68
The `emotion` parameter is deprecated and will be removed in a future version.
""" """
language: Optional[Language] = Language.EN language: Optional[Language] = Language.EN
voice_id: str = "Hades" ## QUESTION: How to make this modifyable/how to modify?
# QUESTION: What about speed, pitch, and temperature??
def __init__( def __init__(
self, self,
@@ -179,6 +178,7 @@ class InworldHttpTTSService(TTSService):
api_key: str, api_key: str,
aiohttp_session: aiohttp.ClientSession, aiohttp_session: aiohttp.ClientSession,
voice_id: str = "Ashley", voice_id: str = "Ashley",
# language: Optional[Language] = Language.EN,
model: str = "inworld-tts-1", model: str = "inworld-tts-1",
base_url: str = "https://api.inworld.ai/tts/v1/voice:stream", base_url: str = "https://api.inworld.ai/tts/v1/voice:stream",
sample_rate: Optional[int] = 48000, sample_rate: Optional[int] = 48000,
@@ -305,6 +305,7 @@ class InworldHttpTTSService(TTSService):
4. Yields audio frames for immediate playback in the pipeline 4. Yields audio frames for immediate playback in the pipeline
Technical Details: Technical Details:
- Uses HTTP streaming with JSON line-by-line responses - Uses HTTP streaming with JSON line-by-line responses
- Each JSON line contains base64-encoded audio data - Each JSON line contains base64-encoded audio data
- Implements buffering to handle partial JSON lines - Implements buffering to handle partial JSON lines
@@ -334,7 +335,7 @@ class InworldHttpTTSService(TTSService):
"audio_config": self._settings[ "audio_config": self._settings[
"audio_config" "audio_config"
], # Audio format settings (LINEAR16, 48kHz) ], # Audio format settings (LINEAR16, 48kHz)
"language": self._settings["language"], # Language code (en, es, etc.) # "language": self._settings["language"], # Language code (en, es, etc.)
} }
# Set up HTTP headers for authentication and content type # Set up HTTP headers for authentication and content type