diff --git a/codes/models/audio/tts/unified_voice2.py b/codes/models/audio/tts/unified_voice2.py index fa85ffc6..435526c6 100644 --- a/codes/models/audio/tts/unified_voice2.py +++ b/codes/models/audio/tts/unified_voice2.py @@ -32,17 +32,17 @@ class ResBlock(nn.Module): class GPT2InferenceModel(GPT2PreTrainedModel): - def __init__(self, config, gpt, posterior_pos_emb, embeddings, norm, linear): + def __init__(self, config, gpt, text_pos_emb, embeddings, norm, linear): super().__init__(config) self.transformer = gpt - self.posterior_pos_embedding = posterior_pos_emb + self.text_pos_embedding = text_pos_emb self.embeddings = embeddings - self.head = nn.Sequential(norm, linear) + self.lm_head = nn.Sequential(norm, linear) # Model parallel self.model_parallel = False self.device_map = None - self.cached_prior_emb = None + self.cached_mel_emb = None def parallelize(self, device_map=None): self.device_map = ( @@ -52,26 +52,27 @@ class GPT2InferenceModel(GPT2PreTrainedModel): ) assert_device_map(self.device_map, len(self.transformer.h)) self.transformer.parallelize(self.device_map) - self.head = self.head.to(self.transformer.first_device) + self.lm_head = self.lm_head.to(self.transformer.first_device) self.model_parallel = True def deparallelize(self): self.transformer.deparallelize() self.transformer = self.transformer.to("cpu") - self.head = self.head.to("cpu") + self.lm_head = self.lm_head.to("cpu") self.model_parallel = False torch.cuda.empty_cache() def get_output_embeddings(self): - return self.head + return self.lm_head def set_output_embeddings(self, new_embeddings): - self.head = new_embeddings + self.lm_head = new_embeddings - def store_prior_emb(self, mel_emb): - self.cached_prior_emb = mel_emb + def store_mel_emb(self, mel_emb): + self.cached_mel_emb = mel_emb def prepare_inputs_for_generation(self, input_ids, past=None, **kwargs): + token_type_ids = kwargs.get("token_type_ids", None) # only last token for inputs_ids if past is defined in kwargs if past: @@ -116,25 +117,25 @@ class GPT2InferenceModel(GPT2PreTrainedModel): output_hidden_states=None, return_dict=None, ): - assert self.cached_prior_emb is not None + assert self.cached_mel_emb is not None assert inputs_embeds is None # Not supported by this inference model. assert labels is None # Training not supported by this inference model. return_dict = return_dict if return_dict is not None else self.config.use_return_dict # Create embedding - prior_len = self.cached_prior_emb.shape[1] + mel_len = self.cached_mel_emb.shape[1] if input_ids.shape[1] != 1: - posterior_inputs = input_ids[:, prior_len:] - posterior_emb = self.embeddings(posterior_inputs) - posterior_emb = posterior_emb + self.posterior_pos_embedding(posterior_emb) - if self.cached_prior_emb.shape[0] != posterior_emb.shape[0]: - prior_emb = self.cached_prior_emb.repeat_interleave(posterior_emb.shape[0] // self.cached_prior_emb.shape[0], 0) + text_inputs = input_ids[:, mel_len:] + text_emb = self.embeddings(text_inputs) + text_emb = text_emb + self.text_pos_embedding(text_emb) + if self.cached_mel_emb.shape[0] != text_emb.shape[0]: + mel_emb = self.cached_mel_emb.repeat_interleave(text_emb.shape[0]//self.cached_mel_emb.shape[0], 0) else: - prior_emb = self.cached_prior_emb - emb = torch.cat([prior_emb, posterior_emb], dim=1) + mel_emb = self.cached_mel_emb + emb = torch.cat([mel_emb, text_emb], dim=1) else: emb = self.embeddings(input_ids) - emb = emb + self.posterior_pos_embedding.get_fixed_embedding(attention_mask.shape[1] - prior_len, attention_mask.device) + emb = emb + self.text_pos_embedding.get_fixed_embedding(attention_mask.shape[1]-mel_len, attention_mask.device) transformer_outputs = self.transformer( inputs_embeds=emb, @@ -155,16 +156,16 @@ class GPT2InferenceModel(GPT2PreTrainedModel): # Set device for model parallelism if self.model_parallel: torch.cuda.set_device(self.transformer.first_device) - hidden_states = hidden_states.to(self.head.weight.device) + hidden_states = hidden_states.to(self.lm_head.weight.device) - logits = self.head(hidden_states) + lm_logits = self.lm_head(hidden_states) if not return_dict: - return (logits,) + transformer_outputs[1:] + return (lm_logits,) + transformer_outputs[1:] return CausalLMOutputWithCrossAttentions( loss=None, - logits=logits, + logits=lm_logits, past_key_values=transformer_outputs.past_key_values, hidden_states=transformer_outputs.hidden_states, attentions=transformer_outputs.attentions, @@ -237,13 +238,35 @@ class MelEncoder(nn.Module): class UnifiedVoice(nn.Module): def __init__(self, layers=8, model_dim=512, heads=8, max_text_tokens=120, max_mel_tokens=250, max_conditioning_inputs=1, - mel_length_compression=1024, number_text_tokens=256, number_mel_codes=8194, start_mel_token=8192, - stop_mel_token=8193, start_text_token=255, checkpointing=True, types=1, only_alignment_head=False): + mel_length_compression=1024, number_text_tokens=256, + start_text_token=255, stop_text_token=0, number_mel_codes=8194, start_mel_token=8192, + stop_mel_token=8193, train_solo_embeddings=False, use_mel_codes_as_input=True, + checkpointing=True, average_conditioning_embeddings=False, freeze_everything_but_position_embeddings=False): + """ + Args: + layers: Number of layers in transformer stack. + model_dim: Operating dimensions of the transformer + heads: Number of transformer heads. Must be divisible by model_dim. Recommend model_dim//64 + max_text_tokens: Maximum number of text tokens that will be encountered by model. + max_mel_tokens: Maximum number of MEL tokens that will be encountered by model. + max_conditioning_inputs: Maximum number of conditioning inputs provided to the model. If (1), conditioning input can be of format (b,80,s), otherwise (b,n,80,s). + mel_length_compression: The factor between and . Used to compute MEL code padding given wav input length. + number_text_tokens: + start_text_token: + stop_text_token: + number_mel_codes: + start_mel_token: + stop_mel_token: + train_solo_embeddings: + use_mel_codes_as_input: + checkpointing: + average_conditioning_embeddings: Whether or not conditioning embeddings should be averaged, instead of fed piecewise into the model. + """ super().__init__() self.number_text_tokens = number_text_tokens - self.start_text_token = number_text_tokens * types if start_text_token is None else start_text_token - self.stop_text_token = 0 + self.start_text_token = start_text_token + self.stop_text_token = stop_text_token self.number_mel_codes = number_mel_codes self.start_mel_token = start_mel_token self.stop_mel_token = stop_mel_token @@ -255,29 +278,41 @@ class UnifiedVoice(nn.Module): self.model_dim = model_dim self.mel_length_compression = mel_length_compression self.conditioning_encoder = ConditioningEncoder(80, model_dim, num_attn_heads=heads) - self.text_embedding = nn.Embedding(self.number_text_tokens*types+1, model_dim) - self.mel_embedding = nn.Embedding(self.number_mel_codes, model_dim) + self.average_conditioning_embeddings = average_conditioning_embeddings + self.text_embedding = nn.Embedding(self.number_text_tokens, model_dim) + if use_mel_codes_as_input: + self.mel_embedding = nn.Embedding(self.number_mel_codes, model_dim) + else: + self.mel_embedding = MelEncoder(model_dim, resblocks_per_reduction=1) self.gpt, self.mel_pos_embedding, self.text_pos_embedding, self.mel_layer_pos_embedding, self.text_layer_pos_embedding = \ build_hf_gpt_transformer(layers, model_dim, heads, self.max_mel_tokens, self.max_text_tokens, checkpointing) + if train_solo_embeddings: + self.mel_solo_embedding = nn.Parameter(torch.randn(1, 1, model_dim) * .02, requires_grad=True) + self.text_solo_embedding = nn.Parameter(torch.randn(1, 1, model_dim) * .02, requires_grad=True) + else: + self.mel_solo_embedding = 0 + self.text_solo_embedding = 0 self.final_norm = nn.LayerNorm(model_dim) - self.text_head = nn.Linear(model_dim, self.number_text_tokens*types+1) + self.text_head = nn.Linear(model_dim, self.number_text_tokens) self.mel_head = nn.Linear(model_dim, self.number_mel_codes) - self.alignment_head = nn.Linear(model_dim, 256) - - if only_alignment_head: - for p in self.parameters(): - p.DO_NOT_TRAIN = True - p.requires_grad = False - for p in self.alignment_head.parameters(): - del p.DO_NOT_TRAIN - p.requires_grad = True # Initialize the embeddings per the GPT-2 scheme - embeddings = [self.text_embedding, self.mel_embedding] + embeddings = [self.text_embedding] + if use_mel_codes_as_input: + embeddings.append(self.mel_embedding) for module in embeddings: module.weight.data.normal_(mean=0.0, std=.02) + if freeze_everything_but_position_embeddings: + for p in self.parameters(): + p.requires_grad = False + p.DO_NOT_TRAIN = True + for m in [self.mel_pos_embedding, self.text_pos_embedding]: + for p in m.parameters(): + del p.DO_NOT_TRAIN + p.requires_grad = True + def get_grad_norm_parameter_groups(self): return { 'conditioning_encoder': list(self.conditioning_encoder.parameters()), @@ -304,43 +339,35 @@ class UnifiedVoice(nn.Module): mel_input_tokens[b, actual_end:] = self.stop_mel_token return mel_input_tokens - def get_logits(self, speech_conditioning_inputs, first_inputs, second_inputs, return_latent=False): - emb = torch.cat([speech_conditioning_inputs, first_inputs, second_inputs], dim=1) + def get_logits(self, speech_conditioning_inputs, first_inputs, first_head, second_inputs=None, second_head=None, get_attns=False, return_latent=False): + if second_inputs is not None: + emb = torch.cat([speech_conditioning_inputs, first_inputs, second_inputs], dim=1) + else: + emb = torch.cat([speech_conditioning_inputs, first_inputs], dim=1) - gpt_out = self.gpt(inputs_embeds=emb, return_dict=True) + gpt_out = self.gpt(inputs_embeds=emb, return_dict=True, output_attentions=get_attns) + if get_attns: + return gpt_out.attentions enc = gpt_out.last_hidden_state[:, 1:] # The first logit is tied to the speech_conditioning_input enc = self.final_norm(enc) if return_latent: - return enc[:, speech_conditioning_inputs.shape[1]:speech_conditioning_inputs.shape[1]+first_inputs.shape[1]], enc[:, -second_inputs.shape[1]:] + return enc[:, :first_inputs.shape[1]], enc[:, -second_inputs.shape[1]:] - text_logits = enc[:, :first_inputs.shape[1]] - text_logits = self.text_head(text_logits) - text_logits = text_logits.permute(0,2,1) + first_logits = enc[:, :first_inputs.shape[1]] + first_logits = first_head(first_logits) + first_logits = first_logits.permute(0,2,1) + if second_inputs is not None: + second_logits = enc[:, -second_inputs.shape[1]:] + second_logits = second_head(second_logits) + second_logits = second_logits.permute(0,2,1) + return first_logits, second_logits + else: + return first_logits - mel_logits = enc[:, -second_inputs.shape[1]:] - mel_logits = self.mel_head(mel_logits) - mel_logits = mel_logits.permute(0,2,1) - - alignment_logits = enc[:, -second_inputs.shape[1]:] - alignment_logits = self.alignment_head(alignment_logits) - alignment_logits = alignment_logits.permute(0,2,1) - - return text_logits, mel_logits, alignment_logits - - - def get_conditioning_latent(self, speech_conditioning_input): - speech_conditioning_input = speech_conditioning_input.unsqueeze(1) if len(speech_conditioning_input.shape) == 3 else speech_conditioning_input - conds = [] - for j in range(speech_conditioning_input.shape[1]): - conds.append(self.conditioning_encoder(speech_conditioning_input[:, j])) - conds = torch.stack(conds, dim=1) - conds = conds.mean(dim=1).unsqueeze(1) - return conds - - - def forward(self, speech_conditioning_input, text_inputs, text_lengths, mel_codes, ctc_codes, wav_lengths, types=None, return_latent=False): + def forward(self, speech_conditioning_input, text_inputs, text_lengths, mel_codes, wav_lengths, text_first=True, raw_mels=None, return_attentions=False, + return_latent=False): """ Forward pass that uses both text and voice in either text conditioning mode or voice conditioning mode (actuated by `text_first`). @@ -350,46 +377,112 @@ class UnifiedVoice(nn.Module): text_lengths: long tensor, (b,) mel_inputs: long tensor, (b,m) wav_lengths: long tensor, (b,) + raw_mels: MEL float tensor (b,80,s) + If return_attentions is specified, only logits are returned. If return_latent is specified, loss & logits are not computed or returned. Only the predicted latents are returned. """ - # Types are expressed by expanding the text embedding space. - if types is not None: - text_inputs = text_inputs * (1+types).unsqueeze(-1) - - # TODO: do this in the dataloader. - for b in range(ctc_codes.shape[0]): - last_code = 0 - for j in range(ctc_codes.shape[1]): - if ctc_codes[b][j] == 0: - ctc_codes[b][j] = last_code - else: - last_code = ctc_codes[b][j] - alignment_targets = F.interpolate(ctc_codes.unsqueeze(1).float(), size=(mel_codes.shape[-1],), mode='nearest').long().squeeze() - + # This model will receive micro-batches with a ton of padding for both the text and MELs. Ameliorate this by + # chopping the inputs by the maximum actual length. + max_text_len = text_lengths.max() + text_inputs = F.pad(text_inputs[:, :max_text_len], (0,1), value=self.stop_text_token) + max_mel_len = wav_lengths.max() // self.mel_length_compression + mel_codes = F.pad(mel_codes[:, :max_mel_len], (0,1), value=self.stop_mel_token) + if raw_mels is not None: + raw_mels = raw_mels[:, :, :max_mel_len*4] mel_codes = self.set_mel_padding(mel_codes, wav_lengths) - text_inputs = F.pad(text_inputs, (0,1), value=self.stop_text_token) - mel_codes = F.pad(mel_codes, (0,1), value=self.stop_mel_token) - alignment_targets = F.pad(alignment_targets, (0,2), value=0) - conds = self.get_conditioning_latent(speech_conditioning_input) + speech_conditioning_input = speech_conditioning_input.unsqueeze(1) if len(speech_conditioning_input.shape) == 3 else speech_conditioning_input + conds = [] + for j in range(speech_conditioning_input.shape[1]): + conds.append(self.conditioning_encoder(speech_conditioning_input[:, j])) + conds = torch.stack(conds, dim=1) + if self.average_conditioning_embeddings: + conds = conds.mean(dim=1).unsqueeze(1) + text_inputs, text_targets = self.build_aligned_inputs_and_targets(text_inputs, self.start_text_token, self.stop_text_token) text_emb = self.text_embedding(text_inputs) + self.text_pos_embedding(text_inputs) mel_codes, mel_targets = self.build_aligned_inputs_and_targets(mel_codes, self.start_mel_token, self.stop_mel_token) - mel_inp = mel_codes + if raw_mels is not None: + mel_inp = F.pad(raw_mels, (0, 8)) + else: + mel_inp = mel_codes mel_emb = self.mel_embedding(mel_inp) mel_emb = mel_emb + self.mel_pos_embedding(mel_codes) - text_logits, mel_logits, alignment_logits = self.get_logits(conds, text_emb, mel_emb, return_latent=return_latent) - if return_latent: - return mel_logits[:, :-2] # Despite the name, these are not logits. Strip off the two tokens added by this forward pass. + if text_first: + text_logits, mel_logits = self.get_logits(conds, text_emb, self.text_head, mel_emb, self.mel_head, get_attns=return_attentions, return_latent=return_latent) + if return_latent: + return mel_logits[:, :-1] # Despite the name, these are not logits. + else: + mel_logits, text_logits = self.get_logits(conds, mel_emb, self.mel_head, text_emb, self.text_head, get_attns=return_attentions, return_latent=return_latent) + if return_latent: + return text_logits[:, :-1] # Despite the name, these are not logits + if return_attentions: + return mel_logits loss_text = F.cross_entropy(text_logits, text_targets.long()) loss_mel = F.cross_entropy(mel_logits, mel_targets.long()) - loss_alignment = F.cross_entropy(alignment_logits, alignment_targets) - return loss_text.mean(), loss_mel.mean(), loss_alignment, mel_logits + return loss_text.mean(), loss_mel.mean(), mel_logits - def inference_speech(self, speech_conditioning_input, text_inputs, **hf_generate_kwargs): + def text_forward(self, speech_conditioning_input, text_inputs, text_lengths): + """ + Performs autoregressive modeling on only text. Still requires a speech_conditioning_input due to the way the + model inputs are formatted. Just provide any audio clip (arguably, zeros could be provided). + """ + # This model will receive micro-batches with a ton of padding for both the text and MELs. Ameliorate this by + # chopping the inputs by the maximum actual length. + max_text_len = text_lengths.max() + text_inputs = F.pad(text_inputs[:, :max_text_len], (0,1), value=self.stop_text_token) + + speech_conditioning_input = speech_conditioning_input.unsqueeze(1) if len(speech_conditioning_input.shape) == 3 else speech_conditioning_input + conds = [] + for j in range(speech_conditioning_input.shape[1]): + conds.append(self.conditioning_encoder(speech_conditioning_input[:, j])) + conds = torch.stack(conds, dim=1) + if self.average_conditioning_embeddings: + conds = conds.mean(dim=1).unsqueeze(1) + + text_inputs, text_targets = self.build_aligned_inputs_and_targets(text_inputs, self.start_text_token, self.stop_text_token) + text_emb = self.text_embedding(text_inputs) + self.text_pos_embedding(text_inputs) + self.text_solo_embedding + text_logits = self.get_logits(conds, text_emb, self.text_head) + loss_text = F.cross_entropy(text_logits, text_targets.long()) + return loss_text.mean() + + def speech_forward(self, speech_conditioning_input, mel_codes, wav_lengths, raw_mels=None): + """ + Performs autoregressive modeling on only speech data. + """ + assert self.max_mel_tokens >= mel_codes.shape[1], f'{mel_codes.shape[1]}' + + # This model will receive micro-batches with a ton of padding for both the text and MELs. Ameliorate this by + # chopping the inputs by the maximum actual length. + max_mel_len = wav_lengths.max() // self.mel_length_compression + mel_codes = F.pad(mel_codes[:, :max_mel_len], (0,1), value=self.stop_mel_token) + mel_codes = self.set_mel_padding(mel_codes, wav_lengths) + if raw_mels is not None: + raw_mels = raw_mels[:, :, :max_mel_len*4] + + speech_conditioning_input = speech_conditioning_input.unsqueeze(1) if len(speech_conditioning_input.shape) == 3 else speech_conditioning_input + conds = [] + for j in range(speech_conditioning_input.shape[1]): + conds.append(self.conditioning_encoder(speech_conditioning_input[:, j])) + conds = torch.stack(conds, dim=1) + if self.average_conditioning_embeddings: + conds = conds.mean(dim=1).unsqueeze(1) + + mel_codes, mel_targets = self.build_aligned_inputs_and_targets(mel_codes, self.start_mel_token, self.stop_mel_token) + if raw_mels is not None: + mel_inp = F.pad(raw_mels, (0, 4)) + else: + mel_inp = mel_codes + mel_emb = self.mel_embedding(mel_inp) + mel_emb = mel_emb + self.mel_pos_embedding(mel_codes) + self.mel_solo_embedding + mel_logits = self.get_logits(conds, mel_emb, self.mel_head) + loss_mel = F.cross_entropy(mel_logits, mel_targets.long()) + return loss_mel.mean() + + def inference_speech(self, speech_conditioning_input, text_inputs, return_attentions=False, **hf_generate_kwargs): if self.max_mel_tokens == -1: # Assume if this is the case, max_mel_tokens=-1 also seq_length = 2002 # Arbitrary default. else: @@ -416,17 +509,64 @@ class UnifiedVoice(nn.Module): for j in range(speech_conditioning_input.shape[1]): conds.append(self.conditioning_encoder(speech_conditioning_input[:, j])) conds = torch.stack(conds, dim=1) - conds = conds.mean(dim=1).unsqueeze(1) + if self.average_conditioning_embeddings: + conds = conds.mean(dim=1).unsqueeze(1) emb = torch.cat([conds, text_emb], dim=1) - self.inference_model.store_prior_emb(emb) + self.inference_model.store_mel_emb(emb) fake_inputs = torch.full((emb.shape[0], conds.shape[1]+emb.shape[1],), fill_value=1, dtype=torch.long, device=text_inputs.device) fake_inputs[:,-1] = self.start_mel_token gen = self.inference_model.generate(fake_inputs, bos_token_id=self.start_mel_token, pad_token_id=self.stop_mel_token, eos_token_id=self.stop_mel_token, - max_length=seq_length, return_dict_in_generate=True, **hf_generate_kwargs) - return gen.sequences[:, fake_inputs.shape[1]:] + max_length=seq_length, output_attentions=return_attentions, return_dict_in_generate=True, **hf_generate_kwargs) + if return_attentions: + return gen.sequences[:, fake_inputs.shape[1]:], gen.attentions + else: + return gen.sequences[:, fake_inputs.shape[1]:] + + + # Turns the (utterly insane) output of HF.generate() into a far more sane output: + # [tensors(B,H,S,S)]. Outer=layers, B=batch,H=head,S=sequence + def make_hf_generate_attentions_sane(self, attentions): + layers = [[] for _ in range(len(attentions[0]))] + full_attention_size = attentions[-1][0].shape[-1] + for i, gen in enumerate(attentions): + for j, lyr in enumerate(gen): + layers[j].append(F.pad(lyr, (0, full_attention_size - lyr.shape[-1]))) + catted = [] + for lyr in layers: + catted.append(torch.cat(lyr, dim=2)) + return catted + + def convert_attentions_to_aligned_codes(self, text, attentions, codes, num_conds): + """ + This was an attempt to make some sense out of the attention matrix retrieved from the unified_voice model. Unfortunately, I can't use it for aligning text & voice. + """ + text_padding = num_conds+2 + num_text = text.shape[-1] + num_context = num_text + text_padding + assert num_context + 1 == attentions[0][0].shape[-1] + attentions = self.make_hf_generate_attentions_sane(attentions) + results = [torch.empty_like(codes) for _ in range(len(attentions))] + for l, layer in enumerate(attentions): + dec_context = layer[:, :, num_context:, :] + # Mask out everything that isn't text (including the start token, which gets a LOT of attention) + dec_context[:,:,:,:text_padding+1] = 0 + dec_context[:,:,:,num_context:] = 0 + for h in range(dec_context.shape[1]): + dec_context_indices = torch.argmax(dec_context[0,h], dim=-1) + print(f'layer_{l};head_{h}: ' + str(dec_context_indices)) + for t, att_tok in enumerate(attentions): + combined_attention_weights = torch.zeros((codes.shape[0], num_text), device=codes.device) + for lyr in att_tok: + token_to_text_attentions = lyr[:, :, -1, text_padding:(text_padding + num_text)].sum(dim=1) + combined_attention_weights = combined_attention_weights + token_to_text_attentions + break + most_attended_text_token = combined_attention_weights.argmax(dim=-1) + results[:, t] = most_attended_text_token + eos_token_mask = (codes != self.stop_mel_token) + return results * eos_token_mask @register_model @@ -435,10 +575,10 @@ def register_unified_voice2(opt_net, opt): if __name__ == '__main__': - gpt = UnifiedVoice(model_dim=256, heads=4, max_conditioning_inputs=4, types=2) + gpt = UnifiedVoice(model_dim=256, heads=4, train_solo_embeddings=True, use_mel_codes_as_input=True, max_conditioning_inputs=4, freeze_everything_but_position_embeddings=True) l = gpt(torch.randn(2, 3, 80, 800), torch.randint(high=256, size=(2,120)), torch.tensor([32, 120]), torch.randint(high=8192, size=(2,250)), - torch.tensor([250*256,195*256]), - types=torch.tensor([0, 1])) + torch.tensor([250*256,195*256])) + #gpt.text_forward(torch.randn(2,80,800), torch.randint(high=50, size=(2,80)), torch.tensor([32, 80])) diff --git a/codes/models/audio/tts/unified_voice4.py b/codes/models/audio/tts/unified_voice4.py new file mode 100644 index 00000000..a186260d --- /dev/null +++ b/codes/models/audio/tts/unified_voice4.py @@ -0,0 +1,444 @@ +import torch +import torch.nn as nn +import torch.nn.functional as F +from transformers import GPT2Config, GPT2PreTrainedModel +from transformers.modeling_outputs import CausalLMOutputWithCrossAttentions +from transformers.models.gpt2.modeling_gpt2 import GPT2Attention +from transformers.utils.model_parallel_utils import get_device_map, assert_device_map + +from models.arch_util import AttentionBlock +from models.audio.tts.transformer_builders import build_hf_gpt_transformer +from models.lucidrains.x_transformers import RotaryEmbedding, apply_rotary_pos_emb +from trainer.networks import register_model +from utils.util import opt_get + + +class ResBlock(nn.Module): + """ + Basic residual convolutional block that uses GroupNorm. + """ + def __init__(self, chan): + super().__init__() + self.net = nn.Sequential( + nn.Conv1d(chan, chan, kernel_size=3, padding=1), + nn.GroupNorm(chan//8, chan), + nn.ReLU(), + nn.Conv1d(chan, chan, kernel_size=3, padding=1), + nn.GroupNorm(chan//8, chan) + ) + + def forward(self, x): + return F.relu(self.net(x) + x) + + +class GPT2InferenceModel(GPT2PreTrainedModel): + def __init__(self, config, gpt, posterior_pos_emb, embeddings, norm, linear): + super().__init__(config) + self.transformer = gpt + self.posterior_pos_embedding = posterior_pos_emb + self.embeddings = embeddings + self.head = nn.Sequential(norm, linear) + + # Model parallel + self.model_parallel = False + self.device_map = None + self.cached_prior_emb = None + + def parallelize(self, device_map=None): + self.device_map = ( + get_device_map(len(self.transformer.h), range(torch.cuda.device_count())) + if device_map is None + else device_map + ) + assert_device_map(self.device_map, len(self.transformer.h)) + self.transformer.parallelize(self.device_map) + self.head = self.head.to(self.transformer.first_device) + self.model_parallel = True + + def deparallelize(self): + self.transformer.deparallelize() + self.transformer = self.transformer.to("cpu") + self.head = self.head.to("cpu") + self.model_parallel = False + torch.cuda.empty_cache() + + def get_output_embeddings(self): + return self.head + + def set_output_embeddings(self, new_embeddings): + self.head = new_embeddings + + def store_prior_emb(self, mel_emb): + self.cached_prior_emb = mel_emb + + def prepare_inputs_for_generation(self, input_ids, past=None, **kwargs): + token_type_ids = kwargs.get("token_type_ids", None) + # only last token for inputs_ids if past is defined in kwargs + if past: + input_ids = input_ids[:, -1].unsqueeze(-1) + if token_type_ids is not None: + token_type_ids = token_type_ids[:, -1].unsqueeze(-1) + + attention_mask = kwargs.get("attention_mask", None) + position_ids = kwargs.get("position_ids", None) + + if attention_mask is not None and position_ids is None: + # create position_ids on the fly for batch generation + position_ids = attention_mask.long().cumsum(-1) - 1 + position_ids.masked_fill_(attention_mask == 0, 1) + if past: + position_ids = position_ids[:, -1].unsqueeze(-1) + else: + position_ids = None + return { + "input_ids": input_ids, + "past_key_values": past, + "use_cache": kwargs.get("use_cache"), + "position_ids": position_ids, + "attention_mask": attention_mask, + "token_type_ids": token_type_ids, + } + + def forward( + self, + input_ids=None, + past_key_values=None, + attention_mask=None, + token_type_ids=None, + position_ids=None, + head_mask=None, + inputs_embeds=None, + encoder_hidden_states=None, + encoder_attention_mask=None, + labels=None, + use_cache=None, + output_attentions=None, + output_hidden_states=None, + return_dict=None, + ): + assert self.cached_prior_emb is not None + assert inputs_embeds is None # Not supported by this inference model. + assert labels is None # Training not supported by this inference model. + return_dict = return_dict if return_dict is not None else self.config.use_return_dict + + # Create embedding + prior_len = self.cached_prior_emb.shape[1] + if input_ids.shape[1] != 1: + posterior_inputs = input_ids[:, prior_len:] + posterior_emb = self.embeddings(posterior_inputs) + posterior_emb = posterior_emb + self.posterior_pos_embedding(posterior_emb) + if self.cached_prior_emb.shape[0] != posterior_emb.shape[0]: + prior_emb = self.cached_prior_emb.repeat_interleave(posterior_emb.shape[0] // self.cached_prior_emb.shape[0], 0) + else: + prior_emb = self.cached_prior_emb + emb = torch.cat([prior_emb, posterior_emb], dim=1) + else: + emb = self.embeddings(input_ids) + emb = emb + self.posterior_pos_embedding.get_fixed_embedding(attention_mask.shape[1] - prior_len, attention_mask.device) + + transformer_outputs = self.transformer( + inputs_embeds=emb, + past_key_values=past_key_values, + attention_mask=attention_mask, + token_type_ids=token_type_ids, + position_ids=position_ids, + head_mask=head_mask, + encoder_hidden_states=encoder_hidden_states, + encoder_attention_mask=encoder_attention_mask, + use_cache=use_cache, + output_attentions=output_attentions, + output_hidden_states=output_hidden_states, + return_dict=return_dict, + ) + hidden_states = transformer_outputs[0] + + # Set device for model parallelism + if self.model_parallel: + torch.cuda.set_device(self.transformer.first_device) + hidden_states = hidden_states.to(self.head.weight.device) + + logits = self.head(hidden_states) + + if not return_dict: + return (logits,) + transformer_outputs[1:] + + return CausalLMOutputWithCrossAttentions( + loss=None, + logits=logits, + past_key_values=transformer_outputs.past_key_values, + hidden_states=transformer_outputs.hidden_states, + attentions=transformer_outputs.attentions, + cross_attentions=transformer_outputs.cross_attentions, + ) + + @staticmethod + def _reorder_cache(past, beam_idx): + """ + This function is used to re-order the :obj:`past_key_values` cache if + :meth:`~transformers.PreTrainedModel.beam_search` or :meth:`~transformers.PreTrainedModel.beam_sample` is + called. This is required to match :obj:`past_key_values` with the correct beam_idx at every generation step. + """ + return tuple( + tuple(past_state.index_select(0, beam_idx.to(past_state.device)) for past_state in layer_past) + for layer_past in past + ) + + +class ConditioningEncoder(nn.Module): + def __init__(self, + spec_dim, + embedding_dim, + attn_blocks=6, + num_attn_heads=4, + do_checkpointing=False, + mean=False): + super().__init__() + attn = [] + self.init = nn.Conv1d(spec_dim, embedding_dim, kernel_size=1) + for a in range(attn_blocks): + attn.append(AttentionBlock(embedding_dim, num_attn_heads, do_checkpoint=do_checkpointing)) + self.attn = nn.Sequential(*attn) + self.dim = embedding_dim + self.do_checkpointing = do_checkpointing + self.mean = mean + + def forward(self, x): + h = self.init(x) + h = self.attn(h) + if self.mean: + return h.mean(dim=2) + else: + return h[:, :, 0] + + +class MelEncoder(nn.Module): + def __init__(self, channels, mel_channels=80, resblocks_per_reduction=2): + super().__init__() + self.channels = channels + self.encoder = nn.Sequential(nn.Conv1d(mel_channels, channels//4, kernel_size=3, padding=1), + nn.Sequential(*[ResBlock(channels//4) for _ in range(resblocks_per_reduction)]), + nn.Conv1d(channels//4, channels//2, kernel_size=3, stride=2, padding=1), + nn.GroupNorm(channels//16, channels//2), + nn.ReLU(), + nn.Sequential(*[ResBlock(channels//2) for _ in range(resblocks_per_reduction)]), + nn.Conv1d(channels//2, channels, kernel_size=3, stride=2, padding=1), + nn.GroupNorm(channels//8, channels), + nn.ReLU(), + nn.Sequential(*[ResBlock(channels) for _ in range(resblocks_per_reduction)]), + ) + self.reduction = 4 + + + def forward(self, x): + for e in self.encoder: + x = e(x) + return x.permute(0,2,1) + + +class UnifiedVoice(nn.Module): + def __init__(self, layers=8, model_dim=512, heads=8, max_text_tokens=120, max_mel_tokens=250, max_conditioning_inputs=1, + mel_length_compression=1024, number_text_tokens=256, number_mel_codes=8194, start_mel_token=8192, + stop_mel_token=8193, start_text_token=255, checkpointing=True, types=1, only_alignment_head=False): + super().__init__() + + self.number_text_tokens = number_text_tokens + self.start_text_token = number_text_tokens * types if start_text_token is None else start_text_token + self.stop_text_token = 0 + self.number_mel_codes = number_mel_codes + self.start_mel_token = start_mel_token + self.stop_mel_token = stop_mel_token + self.layers = layers + self.heads = heads + self.max_conditioning_inputs = max_conditioning_inputs + self.max_mel_tokens = -1 if max_mel_tokens == -1 else max_mel_tokens+2+self.max_conditioning_inputs + self.max_text_tokens = -1 if max_text_tokens == -1 else max_text_tokens+2 + self.model_dim = model_dim + self.mel_length_compression = mel_length_compression + self.conditioning_encoder = ConditioningEncoder(80, model_dim, num_attn_heads=heads) + self.text_embedding = nn.Embedding(self.number_text_tokens*types+1, model_dim) + self.mel_embedding = nn.Embedding(self.number_mel_codes, model_dim) + self.gpt, self.mel_pos_embedding, self.text_pos_embedding, self.mel_layer_pos_embedding, self.text_layer_pos_embedding = \ + build_hf_gpt_transformer(layers, model_dim, heads, self.max_mel_tokens, self.max_text_tokens, checkpointing) + + self.final_norm = nn.LayerNorm(model_dim) + self.text_head = nn.Linear(model_dim, self.number_text_tokens*types+1) + self.mel_head = nn.Linear(model_dim, self.number_mel_codes) + self.alignment_head = nn.Linear(model_dim, 256) + + if only_alignment_head: + for p in self.parameters(): + p.DO_NOT_TRAIN = True + p.requires_grad = False + for p in self.alignment_head.parameters(): + del p.DO_NOT_TRAIN + p.requires_grad = True + + # Initialize the embeddings per the GPT-2 scheme + embeddings = [self.text_embedding, self.mel_embedding] + for module in embeddings: + module.weight.data.normal_(mean=0.0, std=.02) + + def get_grad_norm_parameter_groups(self): + return { + 'conditioning_encoder': list(self.conditioning_encoder.parameters()), + 'gpt': list(self.gpt.parameters()), + 'heads': list(self.text_head.parameters()) + list(self.mel_head.parameters()), + } + + def build_aligned_inputs_and_targets(self, input, start_token, stop_token): + inp = F.pad(input, (1,0), value=start_token) + tar = F.pad(input, (0,1), value=stop_token) + return inp, tar + + def set_mel_padding(self, mel_input_tokens, wav_lengths): + """ + Given mel tokens that are derived from a padded audio clip and the actual lengths of each batch element in + that audio clip, reformats the tokens with STOP_MEL_TOKEN in place of the zero padding. This is required + preformatting to create a working TTS model. + """ + # Set padding areas within MEL (currently it is coded with the MEL code for ). + mel_lengths = wav_lengths // self.mel_length_compression + for b in range(len(mel_lengths)): + actual_end = mel_lengths[b] + 1 # Due to the convolutional nature of how these tokens are generated, it would be best if the model predicts a token past the actual last token. + if actual_end < mel_input_tokens.shape[-1]: + mel_input_tokens[b, actual_end:] = self.stop_mel_token + return mel_input_tokens + + def get_logits(self, speech_conditioning_inputs, first_inputs, second_inputs, return_latent=False): + emb = torch.cat([speech_conditioning_inputs, first_inputs, second_inputs], dim=1) + + gpt_out = self.gpt(inputs_embeds=emb, return_dict=True) + + enc = gpt_out.last_hidden_state[:, 1:] # The first logit is tied to the speech_conditioning_input + enc = self.final_norm(enc) + + if return_latent: + return enc[:, speech_conditioning_inputs.shape[1]:speech_conditioning_inputs.shape[1]+first_inputs.shape[1]], enc[:, -second_inputs.shape[1]:] + + text_logits = enc[:, :first_inputs.shape[1]] + text_logits = self.text_head(text_logits) + text_logits = text_logits.permute(0,2,1) + + mel_logits = enc[:, -second_inputs.shape[1]:] + mel_logits = self.mel_head(mel_logits) + mel_logits = mel_logits.permute(0,2,1) + + alignment_logits = enc[:, -second_inputs.shape[1]:] + alignment_logits = self.alignment_head(alignment_logits) + alignment_logits = alignment_logits.permute(0,2,1) + + return text_logits, mel_logits, alignment_logits + + + def get_conditioning_latent(self, speech_conditioning_input): + speech_conditioning_input = speech_conditioning_input.unsqueeze(1) if len(speech_conditioning_input.shape) == 3 else speech_conditioning_input + conds = [] + for j in range(speech_conditioning_input.shape[1]): + conds.append(self.conditioning_encoder(speech_conditioning_input[:, j])) + conds = torch.stack(conds, dim=1) + conds = conds.mean(dim=1).unsqueeze(1) + return conds + + + def forward(self, speech_conditioning_input, text_inputs, text_lengths, mel_codes, ctc_codes, wav_lengths, types=None, return_latent=False): + """ + Forward pass that uses both text and voice in either text conditioning mode or voice conditioning mode + (actuated by `text_first`). + + speech_conditioning_input: MEL float tensor, (b,80,s) + text_inputs: long tensor, (b,t) + text_lengths: long tensor, (b,) + mel_inputs: long tensor, (b,m) + wav_lengths: long tensor, (b,) + + If return_latent is specified, loss & logits are not computed or returned. Only the predicted latents are returned. + """ + # Types are expressed by expanding the text embedding space. + if types is not None: + text_inputs = text_inputs * (1+types).unsqueeze(-1) + + # TODO: do this in the dataloader. + for b in range(ctc_codes.shape[0]): + last_code = 0 + for j in range(ctc_codes.shape[1]): + if ctc_codes[b][j] == 0: + ctc_codes[b][j] = last_code + else: + last_code = ctc_codes[b][j] + alignment_targets = F.interpolate(ctc_codes.unsqueeze(1).float(), size=(mel_codes.shape[-1],), mode='nearest').long().squeeze() + + mel_codes = self.set_mel_padding(mel_codes, wav_lengths) + text_inputs = F.pad(text_inputs, (0,1), value=self.stop_text_token) + mel_codes = F.pad(mel_codes, (0,1), value=self.stop_mel_token) + alignment_targets = F.pad(alignment_targets, (0,2), value=0) + + conds = self.get_conditioning_latent(speech_conditioning_input) + text_inputs, text_targets = self.build_aligned_inputs_and_targets(text_inputs, self.start_text_token, self.stop_text_token) + text_emb = self.text_embedding(text_inputs) + self.text_pos_embedding(text_inputs) + mel_codes, mel_targets = self.build_aligned_inputs_and_targets(mel_codes, self.start_mel_token, self.stop_mel_token) + mel_inp = mel_codes + mel_emb = self.mel_embedding(mel_inp) + mel_emb = mel_emb + self.mel_pos_embedding(mel_codes) + + text_logits, mel_logits, alignment_logits = self.get_logits(conds, text_emb, mel_emb, return_latent=return_latent) + if return_latent: + return mel_logits[:, :-2] # Despite the name, these are not logits. Strip off the two tokens added by this forward pass. + + loss_text = F.cross_entropy(text_logits, text_targets.long()) + loss_mel = F.cross_entropy(mel_logits, mel_targets.long()) + loss_alignment = F.cross_entropy(alignment_logits, alignment_targets) + return loss_text.mean(), loss_mel.mean(), loss_alignment, mel_logits + + def inference_speech(self, speech_conditioning_input, text_inputs, **hf_generate_kwargs): + if self.max_mel_tokens == -1: # Assume if this is the case, max_mel_tokens=-1 also + seq_length = 2002 # Arbitrary default. + else: + seq_length = self.max_mel_tokens + self.max_text_tokens + 2 + if not hasattr(self, 'inference_model'): + # TODO: Decouple gpt_config from this inference model. + gpt_config = GPT2Config(vocab_size=self.max_mel_tokens, + n_positions=seq_length, + n_ctx=seq_length, + n_embd=self.model_dim, + n_layer=self.layers, + n_head=self.heads, + gradient_checkpointing=False, + use_cache=True) + self.inference_model = GPT2InferenceModel(gpt_config, self.gpt, self.mel_pos_embedding, self.mel_embedding, self.final_norm, self.mel_head) + self.gpt.wte = self.mel_embedding + + text_inputs = F.pad(text_inputs, (0, 1), value=self.stop_text_token) + text_inputs, text_targets = self.build_aligned_inputs_and_targets(text_inputs, self.start_text_token, self.stop_text_token) + text_emb = self.text_embedding(text_inputs) + self.text_pos_embedding(text_inputs) + + speech_conditioning_input = speech_conditioning_input.unsqueeze(1) if len(speech_conditioning_input.shape) == 3 else speech_conditioning_input + conds = [] + for j in range(speech_conditioning_input.shape[1]): + conds.append(self.conditioning_encoder(speech_conditioning_input[:, j])) + conds = torch.stack(conds, dim=1) + conds = conds.mean(dim=1).unsqueeze(1) + + emb = torch.cat([conds, text_emb], dim=1) + self.inference_model.store_prior_emb(emb) + + fake_inputs = torch.full((emb.shape[0], conds.shape[1]+emb.shape[1],), fill_value=1, dtype=torch.long, device=text_inputs.device) + fake_inputs[:,-1] = self.start_mel_token + + gen = self.inference_model.generate(fake_inputs, bos_token_id=self.start_mel_token, pad_token_id=self.stop_mel_token, eos_token_id=self.stop_mel_token, + max_length=seq_length, return_dict_in_generate=True, **hf_generate_kwargs) + return gen.sequences[:, fake_inputs.shape[1]:] + + +@register_model +def register_unified_voice4(opt_net, opt): + return UnifiedVoice(**opt_get(opt_net, ['kwargs'], {})) + + +if __name__ == '__main__': + gpt = UnifiedVoice(model_dim=256, heads=4, max_conditioning_inputs=4, types=2) + l = gpt(torch.randn(2, 3, 80, 800), + torch.randint(high=256, size=(2,120)), + torch.tensor([32, 120]), + torch.randint(high=8192, size=(2,250)), + torch.tensor([250*256,195*256]), + types=torch.tensor([0, 1]))