`xtts` model guide¶

Overview¶

xtts is a VoiceHub text to speech integration. This page is generated from the model registry and its executable data and training contracts, so the documented support stays aligned with code. Open the xtts Colab notebook.

Quickstart¶

python -m pip install voicehub

Install VoiceHub and the provider extra shown above.
Choose a checkpoint that matches this integration.
Provide an authorized reference.wav and an exact reference transcript when the example requests them.
Generate audio and inspect the returned sample rate and metadata.

from pathlib import Path

from voicehub import AutoModelForTextToSpeech, TTSGenerationConfig

model = AutoModelForTextToSpeech.from_pretrained(
    'coqui/XTTS-v2',
    model_type='xtts',
    device="cuda",
    lazy_load=True,
)
generation_kwargs = {
    "speaker_audio_path": str(REFERENCE_AUDIO),
    "language": "en",
}
output = model.generate(
    "VoiceHub keeps model integrations consistent and easy to extend.",
    generation_config=TTSGenerationConfig(
        seed=42,
        output_file=Path("output.wav"),
    ),
    **generation_kwargs,
)
print(output.file_path, output.sample_rate)

Use only authorized recordings for reference voice, transcription, detection, or evaluation. The example selects a concrete device; verify checkpoint-specific hardware needs and pin an immutable revision before production use.

Supported tasks and capabilities¶

Property	Value
Task	Text to speech
Architecture	`xtts2`
Runtime	`VoiceHub-native`
Capabilities	`text-to-speech`, `voice-cloning`, `multilingual`, `fine-tuning`, `safetensors`, `voicehub-native`, `native-runtime`, `preencoded-code-fine-tuning`, `gpt-fine-tuning`, `restricted-pickle-conversion`
Reusable components	—

Data contract¶

Property	Value
Readiness	`preprocessed`
Data architecture	`hybrid`
Sample rate	22,050 Hz
Contract getter	`get_tts_dataset_spec('xtts')`

Variant	Required fields	One of	Boundary	Other rules
`native-gpt-tokens`	`text_inputs`, `text_lengths`, `audio_codes`, `wav_lengths`	cond_mels / cond_latents	Prepared	—
`native-gpt-waveform`	`text_inputs`, `text_lengths`	wav / audio_values; cond_mels / cond_latents	Prepared	at most one: wav / audio_values; forbidden: audio_codes

Multi-component language-model, diffusion, acoustic, or GAN data. Follow the shared data workflow for manifest loading, audio validation, leakage-safe splits, and model-owned preprocessing.

Checkpoints, provenance, and license¶

Property	Value
Default checkpoint	`coqui/XTTS-v2`
Checkpoint status	Registry default; pin an immutable revision for production and reproducible evidence
Implementation	`voicehub.models.xtts.modeling_xtts.XTTSForTextToSpeech`
Configuration	`voicehub.models.xtts.configuration_xtts.XTTSConfig`
Source provenance	`voicehub/models/xtts/source/SOURCE.json`
License	CPML

XTTS checkpoint terms are separate from the MPL-2.0 runtime source. Commercial use: review required.

The default checkpoint identifies the expected family, not every compatible variant. Confirm the selected checkpoint's revision, access terms, provenance, and license before downloading or redistributing it.

Optimization and training support¶

All public optimizations enter this model through the shared BaseSpeechModel lifecycle. Use available_optimization_passes() to discover the public pass registry, then apply, inspect, serialize, or restore a plan through the common model API. Application remains fail-closed when the active runtime or hardware cannot satisfy a pass.

Training contract¶

Property	Value
Support	`preprocessed`
Family	`composite`
Recipe	`single-phase`
Default phase	`language_model`
Training checkpoint	`coqui/XTTS-v2`
Native training graph	`yes`

Phase	Kind	Components	Required inputs	Loss keys
`language_model`	objective	`model.gpt`	`text_inputs`, `text_lengths`, `audio_codes`, `wav_lengths`	`loss`, `loss_text_ce`, `loss_mel_ce`

Prepare the exact tensors listed in the data contract before this step. Call model.validate_training_support() before constructing a trainer. Follow the shared training workflow for a one-step smoke test, validation, checkpoint resume, optimization, and portable export.

Public API¶

Purpose	Public object
Discover	`get_model_spec('xtts')`
Load and run	`AutoModelForTextToSpeech`
Configure	`XTTSConfig`
Model implementation	`XTTSForTextToSpeech`
Normalized output	`TTSOutput`
Training contract	`get_training_spec('xtts')`
Optimization lifecycle	`available_optimization_passes`, `apply_optimization_plan`, `optimization_manifest`, `restore_optimization_plan`

xtts model guide¶