Loading only one decoder and removing lazy loading

This commit is contained in:
WeberJulian 2023-10-04 07:31:21 -03:00
parent 2ecf84a2c6
commit 0d36dcfd81
1 changed files with 96 additions and 108 deletions

View File

@ -198,13 +198,12 @@ class XttsArgs(Coqpit):
Args: Args:
gpt_batch_size (int): The size of the auto-regressive batch. gpt_batch_size (int): The size of the auto-regressive batch.
enable_redaction (bool, optional): Whether to enable redaction. Defaults to True. enable_redaction (bool, optional): Whether to enable redaction. Defaults to True.
lazy_load (bool, optional): Whether to load models on demand. It reduces VRAM usage. Defaults to False.
kv_cache (bool, optional): Whether to use the kv_cache. Defaults to True. kv_cache (bool, optional): Whether to use the kv_cache. Defaults to True.
gpt_checkpoint (str, optional): The checkpoint for the autoregressive model. Defaults to None. gpt_checkpoint (str, optional): The checkpoint for the autoregressive model. Defaults to None.
clvp_checkpoint (str, optional): The checkpoint for the ConditionalLatentVariablePerseq model. Defaults to None. clvp_checkpoint (str, optional): The checkpoint for the ConditionalLatentVariablePerseq model. Defaults to None.
decoder_checkpoint (str, optional): The checkpoint for the DiffTTS model. Defaults to None. decoder_checkpoint (str, optional): The checkpoint for the DiffTTS model. Defaults to None.
num_chars (int, optional): The maximum number of characters to generate. Defaults to 255. num_chars (int, optional): The maximum number of characters to generate. Defaults to 255.
vocoder (VocType, optional): The vocoder to use for synthesis. Defaults to VocConf.Univnet. use_hifigan (bool, optional): Whether to use hifigan or diffusion + univnet as a decoder. Defaults to True.
For GPT model: For GPT model:
ar_max_audio_tokens (int, optional): The maximum mel tokens for the autoregressive model. Defaults to 604. ar_max_audio_tokens (int, optional): The maximum mel tokens for the autoregressive model. Defaults to 604.
@ -234,12 +233,12 @@ class XttsArgs(Coqpit):
gpt_batch_size: int = 1 gpt_batch_size: int = 1
enable_redaction: bool = False enable_redaction: bool = False
lazy_load: bool = True
kv_cache: bool = True kv_cache: bool = True
gpt_checkpoint: str = None gpt_checkpoint: str = None
clvp_checkpoint: str = None clvp_checkpoint: str = None
decoder_checkpoint: str = None decoder_checkpoint: str = None
num_chars: int = 255 num_chars: int = 255
use_hifigan: bool = True
# XTTS GPT Encoder params # XTTS GPT Encoder params
tokenizer_file: str = "" tokenizer_file: str = ""
@ -297,7 +296,6 @@ class Xtts(BaseTTS):
def __init__(self, config: Coqpit): def __init__(self, config: Coqpit):
super().__init__(config, ap=None, tokenizer=None) super().__init__(config, ap=None, tokenizer=None)
self.lazy_load = self.args.lazy_load
self.mel_stats_path = None self.mel_stats_path = None
self.config = config self.config = config
self.gpt_checkpoint = self.args.gpt_checkpoint self.gpt_checkpoint = self.args.gpt_checkpoint
@ -307,7 +305,6 @@ class Xtts(BaseTTS):
self.tokenizer = VoiceBpeTokenizer() self.tokenizer = VoiceBpeTokenizer()
self.gpt = None self.gpt = None
self.diffusion_decoder = None
self.init_models() self.init_models()
self.register_buffer("mel_stats", torch.ones(80)) self.register_buffer("mel_stats", torch.ones(80))
@ -334,6 +331,8 @@ class Xtts(BaseTTS):
stop_audio_token=self.args.gpt_stop_audio_token, stop_audio_token=self.args.gpt_stop_audio_token,
) )
if self.args.use_hifigan:
self.hifigan_decoder = HifiDecoder( self.hifigan_decoder = HifiDecoder(
input_sample_rate=self.args.input_sample_rate, input_sample_rate=self.args.input_sample_rate,
output_sample_rate=self.args.output_sample_rate, output_sample_rate=self.args.output_sample_rate,
@ -344,6 +343,7 @@ class Xtts(BaseTTS):
cond_d_vector_in_each_upsampling_layer=self.args.cond_d_vector_in_each_upsampling_layer, cond_d_vector_in_each_upsampling_layer=self.args.cond_d_vector_in_each_upsampling_layer,
) )
else:
self.diffusion_decoder = DiffusionTts( self.diffusion_decoder = DiffusionTts(
model_channels=self.args.diff_model_channels, model_channels=self.args.diff_model_channels,
num_layers=self.args.diff_num_layers, num_layers=self.args.diff_num_layers,
@ -357,27 +357,12 @@ class Xtts(BaseTTS):
layer_drop=self.args.diff_layer_drop, layer_drop=self.args.diff_layer_drop,
unconditioned_percentage=self.args.diff_unconditioned_percentage, unconditioned_percentage=self.args.diff_unconditioned_percentage,
) )
self.vocoder = UnivNetGenerator() self.vocoder = UnivNetGenerator()
@property @property
def device(self): def device(self):
return next(self.parameters()).device return next(self.parameters()).device
@contextmanager
def lazy_load_model(self, model):
"""Context to load a model on demand.
Args:
model (nn.Module): The model to be loaded.
"""
if self.lazy_load:
yield model
else:
m = model.to(self.device)
yield m
m = model.cpu()
def get_gpt_cond_latents(self, audio_path: str, length: int = 3): def get_gpt_cond_latents(self, audio_path: str, length: int = 3):
"""Compute the conditioning latents for the GPT model from the given audio. """Compute the conditioning latents for the GPT model from the given audio.
@ -411,8 +396,7 @@ class Xtts(BaseTTS):
) )
diffusion_conds.append(cond_mel) diffusion_conds.append(cond_mel)
diffusion_conds = torch.stack(diffusion_conds, dim=1) diffusion_conds = torch.stack(diffusion_conds, dim=1)
with self.lazy_load_model(self.diffusion_decoder) as diffusion: diffusion_latent = self.diffusion_decoder.get_conditioning(diffusion_conds)
diffusion_latent = diffusion.get_conditioning(diffusion_conds)
return diffusion_latent return diffusion_latent
def get_speaker_embedding( def get_speaker_embedding(
@ -430,10 +414,14 @@ class Xtts(BaseTTS):
audio_path, audio_path,
gpt_cond_len=3, gpt_cond_len=3,
): ):
gpt_cond_latents = self.get_gpt_cond_latents(audio_path, length=gpt_cond_len) # [1, 1024, T] speaker_embedding = None
diffusion_cond_latents = self.get_diffusion_cond_latents(audio_path) diffusion_cond_latents = None
if self.args.use_hifigan:
speaker_embedding = self.get_speaker_embedding(audio_path) speaker_embedding = self.get_speaker_embedding(audio_path)
return gpt_cond_latents.to(self.device), diffusion_cond_latents.to(self.device), speaker_embedding else:
diffusion_cond_latents = self.get_diffusion_cond_latents(audio_path)
gpt_cond_latents = self.get_gpt_cond_latents(audio_path, length=gpt_cond_len) # [1, 1024, T]
return gpt_cond_latents, diffusion_cond_latents, speaker_embedding
def synthesize(self, text, config, speaker_wav, language, **kwargs): def synthesize(self, text, config, speaker_wav, language, **kwargs):
"""Synthesize speech with the given input text. """Synthesize speech with the given input text.
@ -500,7 +488,6 @@ class Xtts(BaseTTS):
cond_free_k=2, cond_free_k=2,
diffusion_temperature=1.0, diffusion_temperature=1.0,
decoder_sampler="ddim", decoder_sampler="ddim",
use_hifigan=True,
**hf_generate_kwargs, **hf_generate_kwargs,
): ):
""" """
@ -579,7 +566,6 @@ class Xtts(BaseTTS):
cond_free_k=cond_free_k, cond_free_k=cond_free_k,
diffusion_temperature=diffusion_temperature, diffusion_temperature=diffusion_temperature,
decoder_sampler=decoder_sampler, decoder_sampler=decoder_sampler,
use_hifigan=use_hifigan,
**hf_generate_kwargs, **hf_generate_kwargs,
) )
@ -614,7 +600,7 @@ class Xtts(BaseTTS):
text_tokens.shape[-1] < self.args.gpt_max_text_tokens text_tokens.shape[-1] < self.args.gpt_max_text_tokens
), " ❗ XTTS can only generate text with a maximum of 400 tokens." ), " ❗ XTTS can only generate text with a maximum of 400 tokens."
if not use_hifigan: if not self.args.use_hifigan:
diffuser = load_discrete_vocoder_diffuser( diffuser = load_discrete_vocoder_diffuser(
desired_diffusion_steps=decoder_iterations, desired_diffusion_steps=decoder_iterations,
cond_free=cond_free, cond_free=cond_free,
@ -623,9 +609,7 @@ class Xtts(BaseTTS):
) )
with torch.no_grad(): with torch.no_grad():
self.gpt = self.gpt.to(self.device) gpt_codes = self.gpt.generate(
with self.lazy_load_model(self.gpt) as gpt:
gpt_codes = gpt.generate(
cond_latents=gpt_cond_latent, cond_latents=gpt_cond_latent,
text_inputs=text_tokens, text_inputs=text_tokens,
input_tokens=None, input_tokens=None,
@ -643,7 +627,7 @@ class Xtts(BaseTTS):
[gpt_codes.shape[-1] * self.gpt.code_stride_len], device=text_tokens.device [gpt_codes.shape[-1] * self.gpt.code_stride_len], device=text_tokens.device
) )
text_len = torch.tensor([text_tokens.shape[-1]], device=self.device) text_len = torch.tensor([text_tokens.shape[-1]], device=self.device)
gpt_latents = gpt( gpt_latents = self.gpt(
text_tokens, text_tokens,
text_len, text_len,
gpt_codes, gpt_codes,
@ -663,20 +647,17 @@ class Xtts(BaseTTS):
gpt_latents = gpt_latents[:, :k] gpt_latents = gpt_latents[:, :k]
break break
if use_hifigan: if self.args.use_hifigan:
with self.lazy_load_model(self.hifigan_decoder) as hifigan_decoder: wav = self.hifigan_decoder(gpt_latents, g=speaker_embedding)
wav = hifigan_decoder(gpt_latents, g=speaker_embedding)
else: else:
with self.lazy_load_model(self.diffusion_decoder) as diffusion:
mel = do_spectrogram_diffusion( mel = do_spectrogram_diffusion(
diffusion, self.diffusion_decoder,
diffuser, diffuser,
gpt_latents, gpt_latents,
diffusion_conditioning, diffusion_conditioning,
temperature=diffusion_temperature, temperature=diffusion_temperature,
) )
with self.lazy_load_model(self.vocoder) as vocoder: wav = self.vocoder.inference(mel)
wav = vocoder.inference(mel)
return {"wav": wav.cpu().numpy().squeeze()} return {"wav": wav.cpu().numpy().squeeze()}
@ -713,6 +694,7 @@ class Xtts(BaseTTS):
# Decoder inference # Decoder inference
**hf_generate_kwargs, **hf_generate_kwargs,
): ):
assert hasattr(self, "hifigan_decoder"), "`inference_stream` requires use_hifigan to be set to true in the config.model_args, diffusion is too slow to stream."
text = f"[{language}]{text.strip().lower()}" text = f"[{language}]{text.strip().lower()}"
text_tokens = torch.IntTensor(self.tokenizer.encode(text, lang=language)).unsqueeze(0).to(self.device) text_tokens = torch.IntTensor(self.tokenizer.encode(text, lang=language)).unsqueeze(0).to(self.device)
@ -781,7 +763,7 @@ class Xtts(BaseTTS):
vocab_path=None, vocab_path=None,
eval=False, eval=False,
strict=True, strict=True,
use_deepspeed=False use_deepspeed=False,
): ):
""" """
Loads a checkpoint from disk and initializes the model's state and tokenizer. Loads a checkpoint from disk and initializes the model's state and tokenizer.
@ -807,14 +789,20 @@ class Xtts(BaseTTS):
self.init_models() self.init_models()
if eval: if eval:
self.gpt.init_gpt_for_inference(kv_cache=self.args.kv_cache) self.gpt.init_gpt_for_inference(kv_cache=self.args.kv_cache)
self.load_state_dict(load_fsspec(model_path, map_location=self.device)["model"], strict=strict)
checkpoint = load_fsspec(model_path, map_location=torch.device("cpu"))["model"]
ignore_keys = ["diffusion_decoder", "vocoder"] if self.args.use_hifigan else ["hifigan_decoder"]
for key in list(checkpoint.keys()):
if key.split(".")[0] in ignore_keys:
del checkpoint[key]
self.load_state_dict(checkpoint, strict=strict)
if eval: if eval:
if hasattr(self, "hifigan_decoder"): self.hifigan_decoder.eval()
if hasattr(self, "diffusion_decoder"): self.diffusion_decoder.eval()
if hasattr(self, "vocoder"): self.vocoder.eval()
self.gpt.init_gpt_for_inference(kv_cache=self.args.kv_cache, use_deepspeed=use_deepspeed) self.gpt.init_gpt_for_inference(kv_cache=self.args.kv_cache, use_deepspeed=use_deepspeed)
self.gpt.eval() self.gpt.eval()
self.diffusion_decoder.eval()
self.vocoder.eval()
self.hifigan_decoder.eval()
def train_step(self): def train_step(self):
raise NotImplementedError("XTTS Training is not implemented") raise NotImplementedError("XTTS Training is not implemented")