diffrhythm/model/cfm.py

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

from __future__ import annotations
from typing import Callable
from random import random

import torch
from torch import nn
import torch
import torch.nn.functional as F
from torch.nn.utils.rnn import pad_sequence

from torchdiffeq import odeint

from diffrhythm.model.modules import MelSpec
from diffrhythm.model.utils import (
    default,
    exists,
    list_str_to_idx,
    list_str_to_tensor,
    lens_to_mask,
    mask_from_frac_lengths,
)

def custom_mask_from_start_end_indices(seq_len: int["b"], start: int["b"], end: int["b"], device, max_seq_len):  # noqa: F722 F821
    max_seq_len = max_seq_len
    seq = torch.arange(max_seq_len, device=device).long()
    start_mask = seq[None, :] >= start[:, None]
    end_mask = seq[None, :] < end[:, None]
    return start_mask & end_mask

class CFM(nn.Module):
    def __init__(
        self,
        transformer: nn.Module,
        sigma=0.0,
        odeint_kwargs: dict = dict(
            # atol = 1e-5,
            # rtol = 1e-5,
            method="euler" # 'midpoint'
            # method="adaptive_heun"  # dopri5
        ),
        odeint_options: dict = dict(
            min_step=0.05
        ),
        audio_drop_prob=0.3,
        cond_drop_prob=0.2,
        style_drop_prob=0.1,
        lrc_drop_prob=0.1,
        num_channels=None,
        frac_lengths_mask: tuple[float, float] = (0.7, 1.0),
        vocab_char_map: dict[str:int] | None = None,
        use_style_prompt: bool = False
    ):
        super().__init__()

        self.frac_lengths_mask = frac_lengths_mask

        self.num_channels = num_channels

        # classifier-free guidance
        self.audio_drop_prob = audio_drop_prob
        self.cond_drop_prob = cond_drop_prob
        self.style_drop_prob = style_drop_prob
        self.lrc_drop_prob = lrc_drop_prob

        print(f"audio drop prob -> {self.audio_drop_prob}; style_drop_prob -> {self.style_drop_prob}; lrc_drop_prob: {self.lrc_drop_prob}")

        # transformer
        self.transformer = transformer
        dim = transformer.dim
        self.dim = dim

        # conditional flow related
        self.sigma = sigma

        # sampling related
        self.odeint_kwargs = odeint_kwargs
        # print(f"ODE SOLVER: {self.odeint_kwargs['method']}")
        
        self.odeint_options = odeint_options

        # vocab map for tokenization
        self.vocab_char_map = vocab_char_map

        self.use_style_prompt = use_style_prompt

    @property
    def device(self):
        return next(self.parameters()).device

    @torch.no_grad()
    def sample(
        self,
        cond: float["b n d"] | float["b nw"],  # noqa: F722
        text: int["b nt"] | list[str],  # noqa: F722
        duration: int | int["b"],  # noqa: F821
        *,
        style_prompt = None,
        style_prompt_lens = None,
        negative_style_prompt = None,
        lens: int["b"] | None = None,  # noqa: F821
        steps=32,
        cfg_strength=4.0,
        sway_sampling_coef=None,
        seed: int | None = None,
        max_duration=4096,
        vocoder: Callable[[float["b d n"]], float["b nw"]] | None = None,  # noqa: F722
        no_ref_audio=False,
        duplicate_test=False,
        t_inter=0.1,
        edit_mask=None,
        start_time=None,
        latent_pred_start_frame=0,
        latent_pred_end_frame=2048,
    ):
        self.eval()

        if next(self.parameters()).dtype == torch.float16:
            cond = cond.half()

        # raw wave
        
        if cond.shape[1] > duration:
            cond = cond[:, :duration, :]

        if cond.ndim == 2:
            cond = self.mel_spec(cond)
            cond = cond.permute(0, 2, 1)
            assert cond.shape[-1] == self.num_channels

        batch, cond_seq_len, device = *cond.shape[:2], cond.device
        if not exists(lens):
            lens = torch.full((batch,), cond_seq_len, device=device, dtype=torch.long)

        # text

        if isinstance(text, list):
            if exists(self.vocab_char_map):
                text = list_str_to_idx(text, self.vocab_char_map).to(device)
            else:
                text = list_str_to_tensor(text).to(device)
            assert text.shape[0] == batch

        if exists(text):
            text_lens = (text != -1).sum(dim=-1)
            #lens = torch.maximum(text_lens, lens)  # make sure lengths are at least those of the text characters

        # duration
        # import pdb; pdb.set_trace()
        cond_mask = lens_to_mask(lens)
        if edit_mask is not None:
            cond_mask = cond_mask & edit_mask

        latent_pred_start_frame = torch.tensor([latent_pred_start_frame]).to(cond.device)
        latent_pred_end_frame = duration
        latent_pred_end_frame = torch.tensor([latent_pred_end_frame]).to(cond.device)
        fixed_span_mask = custom_mask_from_start_end_indices(cond_seq_len, latent_pred_start_frame, latent_pred_end_frame, device=cond.device, max_seq_len=duration)

        fixed_span_mask = fixed_span_mask.unsqueeze(-1)
        step_cond = torch.where(fixed_span_mask, torch.zeros_like(cond), cond)

        if isinstance(duration, int):
            duration = torch.full((batch,), duration, device=device, dtype=torch.long)

        # duration = torch.maximum(lens + 1, duration)  # just add one token so something is generated
        duration = duration.clamp(max=max_duration)
        max_duration = duration.amax()

        # duplicate test corner for inner time step oberservation
        if duplicate_test:
            test_cond = F.pad(cond, (0, 0, cond_seq_len, max_duration - 2 * cond_seq_len), value=0.0)

        # cond = F.pad(cond, (0, 0, 0, max_duration - cond_seq_len), value=0.0) # [b, t, d]
        # cond_mask = F.pad(cond_mask, (0, max_duration - cond_mask.shape[-1]), value=False) # [b, max_duration]
        # cond_mask = cond_mask.unsqueeze(-1) #[b, t, d]
        # step_cond = torch.where(
        #     cond_mask, cond, torch.zeros_like(cond)
        # )  # allow direct control (cut cond audio) with lens passed in

        if batch > 1:
            mask = lens_to_mask(duration)
        else:  # save memory and speed up, as single inference need no mask currently
            mask = None

        # test for no ref audio
        if no_ref_audio:
            cond = torch.zeros_like(cond)
        
        start_time_embed, positive_text_embed, positive_text_residuals = self.transformer.forward_timestep_invariant(text, step_cond.shape[1], drop_text=False, start_time=start_time)
        _, negative_text_embed, negative_text_residuals = self.transformer.forward_timestep_invariant(text, step_cond.shape[1], drop_text=True, start_time=start_time)

        text_embed = torch.cat([positive_text_embed, negative_text_embed], 0)
        text_residuals = [torch.cat([a, b], 0) for a, b in zip(positive_text_residuals, negative_text_residuals)]
        step_cond = torch.cat([step_cond, step_cond], 0)
        style_prompt = torch.cat([style_prompt, negative_style_prompt], 0)
        start_time_embed = torch.cat([start_time_embed, start_time_embed], 0)
            

        def fn(t, x):
            x = torch.cat([x, x], 0)
            pred = self.transformer(
                x=x, text_embed=text_embed, text_residuals=text_residuals, cond=step_cond, time=t, 
                drop_audio_cond=True, drop_prompt=False, style_prompt=style_prompt, start_time=start_time_embed
            )

            positive_pred, negative_pred = pred.chunk(2, 0)
            cfg_pred = positive_pred + (positive_pred - negative_pred) * cfg_strength

            return cfg_pred

        # noise input
        # to make sure batch inference result is same with different batch size, and for sure single inference
        # still some difference maybe due to convolutional layers
        y0 = []
        for dur in duration:
            if exists(seed):
                torch.manual_seed(seed)
            y0.append(torch.randn(dur, self.num_channels, device=self.device, dtype=step_cond.dtype))
        y0 = pad_sequence(y0, padding_value=0, batch_first=True)

        t_start = 0

        # duplicate test corner for inner time step oberservation
        if duplicate_test:
            t_start = t_inter
            y0 = (1 - t_start) * y0 + t_start * test_cond
            steps = int(steps * (1 - t_start))

        t = torch.linspace(t_start, 1, steps, device=self.device, dtype=step_cond.dtype)
        if sway_sampling_coef is not None:
            t = t + sway_sampling_coef * (torch.cos(torch.pi / 2 * t) - 1 + t)

        trajectory = odeint(fn, y0, t, **self.odeint_kwargs)

        sampled = trajectory[-1]
        out = sampled
        # out = torch.where(cond_mask, cond, out)
        out = torch.where(fixed_span_mask, out, cond)

        if exists(vocoder):
            out = out.permute(0, 2, 1)
            out = vocoder(out)

        return out, trajectory

    def forward(
        self,
        inp: float["b n d"] | float["b nw"],  # mel or raw wave  # noqa: F722
        text: int["b nt"] | list[str],  # noqa: F722
        style_prompt = None,
        style_prompt_lens = None,
        lens: int["b"] | None = None,  # noqa: F821
        noise_scheduler: str | None = None,
        grad_ckpt = False,
        start_time = None,
    ):

        batch, seq_len, dtype, device, _σ1 = *inp.shape[:2], inp.dtype, self.device, self.sigma

        # lens and mask
        if not exists(lens):
            lens = torch.full((batch,), seq_len, device=device)

        mask = lens_to_mask(lens, length=seq_len)  # useless here, as collate_fn will pad to max length in batch

        # get a random span to mask out for training conditionally
        frac_lengths = torch.zeros((batch,), device=self.device).float().uniform_(*self.frac_lengths_mask)
        rand_span_mask = mask_from_frac_lengths(lens, frac_lengths)

        if exists(mask):
            rand_span_mask = mask
            # rand_span_mask &= mask

        # mel is x1
        x1 = inp

        # x0 is gaussian noise
        x0 = torch.randn_like(x1)

        # time step
        # time = torch.rand((batch,), dtype=dtype, device=self.device)
        time = torch.normal(mean=0, std=1, size=(batch,), device=self.device)
        time = torch.nn.functional.sigmoid(time)
        # TODO. noise_scheduler

        # sample xt (φ_t(x) in the paper)
        t = time.unsqueeze(-1).unsqueeze(-1)
        φ = (1 - t) * x0 + t * x1
        flow = x1 - x0

        # only predict what is within the random mask span for infilling
        cond = torch.where(rand_span_mask[..., None], torch.zeros_like(x1), x1)

        # transformer and cfg training with a drop rate
        drop_audio_cond = random() < self.audio_drop_prob  # p_drop in voicebox paper
        drop_text = random() < self.lrc_drop_prob
        drop_prompt = random() < self.style_drop_prob

        # if want rigourously mask out padding, record in collate_fn in dataset.py, and pass in here
        # adding mask will use more memory, thus also need to adjust batchsampler with scaled down threshold for long sequences
        pred = self.transformer(
            x=φ, cond=cond, text=text, time=time, drop_audio_cond=drop_audio_cond, drop_text=drop_text, drop_prompt=drop_prompt,
            style_prompt=style_prompt, style_prompt_lens=style_prompt_lens, grad_ckpt=grad_ckpt, start_time=start_time
        )

        # flow matching loss
        loss = F.mse_loss(pred, flow, reduction="none")
        loss = loss[rand_span_mask]

        return loss.mean(), cond, pred