GitHubEdit on GitHub

Basic Usage

Getting Started with Amphion

This guide will walk you through the basic usage of Amphion's core features.

Text-to-Speech

Here's a simple example of text-to-speech synthesis:

from amphion import TextToSpeech

# Initialize the model
tts = TextToSpeech()

# Basic synthesis
audio = tts.synthesize("Hello, world!")
audio.save("output.wav")

# Synthesis with different speaker
audio = tts.synthesize("Hello, world!", speaker="speaker_1")
audio.save("output_speaker1.wav")

# Synthesis with emotion
audio = tts.synthesize("Hello, world!", emotion="happy")
audio.save("output_happy.wav")

Voice Conversion

Convert voice from one speaker to another:

from amphion import VoiceConverter

# Initialize the converter
converter = VoiceConverter()

# Convert voice
converted_audio = converter.convert(
    source="input.wav",
    target_speaker="target_speaker"
)
converted_audio.save("converted.wav")

Text-to-Audio

Generate audio from text descriptions:

from amphion import TextToAudio

# Initialize the generator
generator = TextToAudio()

# Generate audio from description
audio = generator.generate(
    "A calm forest ambience with birds chirping and leaves rustling"
)
audio.save("forest.wav")

Configuration

You can customize the behavior of Amphion components through configuration:

from amphion import TextToSpeech
from amphion.config import Config

config = Config(
    sample_rate=44100,
    device="cuda",
    model_path="path/to/custom/model"
)

tts = TextToSpeech(config)

Error Handling

Amphion provides clear error messages and exceptions:

try:
    audio = tts.synthesize("Hello, world!")
except ModelNotFoundError:
    print("Model not found. Please check your installation.")
except DeviceNotAvailableError:
    print("CUDA device not available. Falling back to CPU.")

Next Steps

InstallationConfiguration