Audio & Voice

from gaia.audio.audio_client import AudioClient

# Initialize audio client
audio = AudioClient(
    whisper_model_size="base",    # ASR model: base, small, medium, large
    audio_device_index=None,       # Auto-select
    silence_threshold=0.05,        # Voice detection sensitivity
    enable_tts=True                # Enable text-to-speech
)

# List audio devices
devices = audio.get_device_list()
for idx, device in enumerate(devices):
    print(f"{idx}: {device}")

# Define message processor
def process_user_message(message: str) -> str:
    # Your agent logic
    return f"You said: {message}"

# Start voice chat
audio.start_voice_chat(message_processor_callback=process_user_message)

from gaia.audio.whisper_asr import WhisperAsr
from queue import Queue

# Create transcription queue
transcription_queue = Queue()

# Initialize Whisper
asr = WhisperAsr(
    model_size="base",
    device_index=0,
    transcription_queue=transcription_queue,
    enable_cuda=False,  # Use CUDA if available
    silence_threshold=0.05
)

# Start recording
asr.start_recording()

# Get transcriptions
while True:
    if not transcription_queue.empty():
        text = transcription_queue.get()
        print(f"Transcribed: {text}")
        # Process text...

# Stop recording
asr.stop_recording()

from gaia.audio.kokoro_tts import KokoroTTS
import wave

# Initialize TTS
tts = KokoroTTS()

# List available voices
voices = tts.list_voices()
for voice_id, voice_data in voices.items():
    print(f"{voice_id}: {voice_data['description']}")

# Generate speech
text = "Hello! This is a test of the Kokoro text-to-speech system."
audio_bytes = tts.synthesize(
    text=text,
    voice="af_sarah"  # American female voice
)

# Save to file
with wave.open("output.wav", "wb") as wav_file:
    wav_file.setnchannels(1)
    wav_file.setsampwidth(2)
    wav_file.setframerate(24000)
    wav_file.writeframes(audio_bytes)

# Or play directly
import sounddevice as sd
import numpy as np
audio_array = np.frombuffer(audio_bytes, dtype=np.int16)
sd.play(audio_array, 24000)
sd.wait()