f5_tts/model/backbones/dit.py

"""
ein notation:
b - batch
n - sequence
nt - text sequence
nw - raw wave length
d - dimension
"""

from __future__ import annotations

import torch
import torch.nn.functional as F
from torch import nn
from x_transformers.x_transformers import RotaryEmbedding

from f5_tts.model.modules import (AdaLayerNormZero_Final, ConvNeXtV2Block,
                                  ConvPositionEmbedding, DiTBlock,
                                  TimestepEmbedding, get_pos_embed_indices,
                                  precompute_freqs_cis)

# Text embedding


class TextEmbedding(nn.Module):
    def __init__(self, text_num_embeds, text_dim, conv_layers=0, conv_mult=2):
        super().__init__()
        self.text_embed = nn.Embedding(
            text_num_embeds + 1, text_dim
        )  # use 0 as filler token

        if conv_layers > 0:
            self.extra_modeling = True
            self.precompute_max_pos = 4096  # ~44s of 24khz audio
            self.register_buffer(
                "freqs_cis",
                precompute_freqs_cis(text_dim, self.precompute_max_pos),
                persistent=False,
            )
            self.text_blocks = nn.Sequential(
                *[
                    ConvNeXtV2Block(text_dim, text_dim * conv_mult)
                    for _ in range(conv_layers)
                ]
            )
        else:
            self.extra_modeling = False

    def forward(self, text: int["b nt"], seq_len, drop_text=False):  # noqa: F722
        text = (
            text + 1
        )  # use 0 as filler token. preprocess of batch pad -1, see list_str_to_idx()
        text = text[
            :, :seq_len
        ]  # curtail if character tokens are more than the mel spec tokens
        batch, text_len = text.shape[0], text.shape[1]
        text = F.pad(text, (0, seq_len - text_len), value=0)

        if drop_text:  # cfg for text
            text = torch.zeros_like(text)

        text = self.text_embed(text)  # b n -> b n d

        # possible extra modeling
        if self.extra_modeling:
            # sinus pos emb
            batch_start = torch.zeros((batch,), dtype=torch.long)
            pos_idx = get_pos_embed_indices(
                batch_start, seq_len, max_pos=self.precompute_max_pos
            )
            text_pos_embed = self.freqs_cis[pos_idx]
            text = text + text_pos_embed

            # convnextv2 blocks
            text = self.text_blocks(text)

        return text


# noised input audio and context mixing embedding


class InputEmbedding(nn.Module):
    def __init__(self, mel_dim, text_dim, out_dim):
        super().__init__()
        self.proj = nn.Linear(mel_dim * 2 + text_dim, out_dim)
        self.conv_pos_embed = ConvPositionEmbedding(dim=out_dim)

    def forward(
        self,
        x: float["b n d"],
        cond: float["b n d"],
        text_embed: float["b n d"],
        drop_audio_cond=False,
    ):  # noqa: F722
        if drop_audio_cond:  # cfg for cond audio
            cond = torch.zeros_like(cond)

        x = self.proj(torch.cat((x, cond, text_embed), dim=-1))
        x = self.conv_pos_embed(x) + x
        return x


# Transformer backbone using DiT blocks


class DiT(nn.Module):
    def __init__(
        self,
        *,
        dim,
        depth=8,
        heads=8,
        dim_head=64,
        dropout=0.1,
        ff_mult=4,
        mel_dim=100,
        text_num_embeds=256,
        text_dim=None,
        conv_layers=0,
        long_skip_connection=False,
        checkpoint_activations=False,
        second_time=False,
    ):
        super().__init__()

        self.time_embed = TimestepEmbedding(dim)
        if second_time:
            self.time_embed2 = TimestepEmbedding(dim)
            # Zero-init the weights and biases of the first and last Linear layers in time_mlp
            nn.init.zeros_(
                self.time_embed2.time_mlp[0].weight
            )  # First Linear layer weights
            nn.init.zeros_(self.time_embed2.time_mlp[0].bias)  # First Linear layer bias
            nn.init.zeros_(
                self.time_embed2.time_mlp[-1].weight
            )  # Last Linear layer weights
            nn.init.zeros_(self.time_embed2.time_mlp[-1].bias)  # Last Linear layer bias
        else:
            self.time_embed2 = None

        if text_dim is None:
            text_dim = mel_dim
        self.vocab_size = text_num_embeds
        self.text_embed = TextEmbedding(
            text_num_embeds, text_dim, conv_layers=conv_layers
        )
        self.input_embed = InputEmbedding(mel_dim, text_dim, dim)

        self.rotary_embed = RotaryEmbedding(dim_head)

        self.dim = dim
        self.depth = depth

        self.transformer_blocks = nn.ModuleList(
            [
                DiTBlock(
                    dim=dim,
                    heads=heads,
                    dim_head=dim_head,
                    ff_mult=ff_mult,
                    dropout=dropout,
                )
                for _ in range(depth)
            ]
        )
        self.long_skip_connection = (
            nn.Linear(dim * 2, dim, bias=False) if long_skip_connection else None
        )

        self.norm_out = AdaLayerNormZero_Final(dim)  # final modulation
        self.proj_out = nn.Linear(dim, mel_dim)

        self.checkpoint_activations = checkpoint_activations

    def ckpt_wrapper(self, module):
        # https://github.com/chuanyangjin/fast-DiT/blob/main/models.py
        def ckpt_forward(*inputs):
            outputs = module(*inputs)
            return outputs

        return ckpt_forward

    def forward(
        self,
        x: float["b n d"],  # nosied input audio  # noqa: F722
        cond: float["b n d"],  # masked cond audio  # noqa: F722
        text: int["b nt"],  # text  # noqa: F722
        time: float["b"] | float[""],  # time step  # noqa: F821 F722
        drop_audio_cond,  # cfg for cond audio
        drop_text,  # cfg for text
        mask: bool["b n"] | None = None,  # noqa: F722
        second_time: float["b"] | float[""] = None,  # noqa: F821 F722
        classify_mode: bool = False,  # noqa: F821
    ):
        batch, seq_len = x.shape[0], x.shape[1]
        if time.ndim == 0:
            time = time.repeat(batch)

        # t: conditioning time, c: context (text + masked cond audio), x: noised input audio
        t = self.time_embed(time)
        if second_time is not None and self.time_embed2 is not None:
            t2 = self.time_embed2(second_time)
            t = t + t2

        text_embed = self.text_embed(text, seq_len, drop_text=drop_text)
        x = self.input_embed(x, cond, text_embed, drop_audio_cond=drop_audio_cond)

        rope = self.rotary_embed.forward_from_seq_len(seq_len)

        if self.long_skip_connection is not None:
            residual = x

        if classify_mode:
            layers = [x]

        for block in self.transformer_blocks:
            if self.checkpoint_activations:
                x = torch.utils.checkpoint.checkpoint(
                    self.ckpt_wrapper(block), x, t, mask, rope
                )
            else:
                x = block(x, t, mask=mask, rope=rope)

            if classify_mode:
                layers.append(x)

        if self.long_skip_connection is not None:
            x = self.long_skip_connection(torch.cat((x, residual), dim=-1))

        x = self.norm_out(x, t)
        output = self.proj_out(x)

        if classify_mode:
            return layers

        return output