Add gpt_segmentor model

The idea is to specifically train a model that extracts phrases from
audio clips.
This commit is contained in:
James Betker 2021-08-15 21:23:07 -06:00
parent a826d5f658
commit 9e47e64d5a

View File

@ -0,0 +1,101 @@
import torch
import torch.nn as nn
import torch.nn.functional as F
from munch import munchify
from models.gpt_voice.lucidrains_gpt import Transformer
from models.tacotron2.taco_utils import get_mask_from_lengths
from models.tacotron2.text import symbols, sequence_to_text
from trainer.networks import register_model
from utils.util import opt_get
class ResBlock(nn.Module):
def __init__(self, chan):
super().__init__()
self.net = nn.Sequential(
nn.Conv1d(chan, chan, kernel_size=5, padding = 2),
nn.BatchNorm1d(chan),
nn.ReLU(),
nn.Conv1d(chan, chan, kernel_size=5, padding = 2),
nn.BatchNorm1d(chan)
)
def forward(self, x):
return F.relu(self.net(x) + x)
class MelEncoder(nn.Module):
def __init__(self, channels, mel_channels=80):
super().__init__()
self.channels = channels
self.encoder = nn.Sequential(nn.Conv1d(mel_channels, channels//4, kernel_size=7, padding=3),
ResBlock(channels//4),
ResBlock(channels//4),
nn.Conv1d(channels//4, channels//2, kernel_size=5, stride=2, padding=2),
nn.BatchNorm1d(channels//2),
nn.ReLU(),
ResBlock(channels//2),
ResBlock(channels//2),
ResBlock(channels//2),
nn.Conv1d(channels//2, channels, kernel_size=5, stride=2, padding=2),
ResBlock(channels),
ResBlock(channels),
ResBlock(channels)
)
def forward(self, x):
return self.encoder(x)
class GptSegmentor(nn.Module):
MAX_SYMBOLS_PER_PHRASE = 200
MAX_MEL_FRAMES = 2000 // 4
def __init__(self, layers=8, model_dim=512, heads=8):
super().__init__()
self.model_dim = model_dim
self.max_mel_frames = self.MAX_MEL_FRAMES
self.mel_encoder = MelEncoder(model_dim)
self.mel_pos_embedding = nn.Embedding(self.MAX_MEL_FRAMES, model_dim)
self.gpt = Transformer(dim=model_dim, depth=layers, seq_len=2+self.MAX_SYMBOLS_PER_PHRASE+self.MAX_MEL_FRAMES, heads=heads,
attn_dropout=.1, ff_dropout=.1, non_causal_sequence_partition=self.MAX_MEL_FRAMES)
self.final_norm = nn.LayerNorm(model_dim)
self.stop_head = nn.Linear(model_dim, 1)
def forward(self, mel_inputs, mel_lengths):
mel_emb = self.mel_encoder(mel_inputs)
mel_lengths = mel_lengths // 4 # The encoder decimates the mel by a factor of 4.
mel_emb = mel_emb.permute(0,2,1).contiguous()
mel_emb = mel_emb + self.mel_pos_embedding(torch.arange(mel_emb.shape[1], device=mel_emb.device))
enc = self.gpt(mel_emb)
# Compute loss
b, s, _ = enc.shape
mel_pad_mask = ~get_mask_from_lengths(mel_lengths-1, s)
targets = torch.zeros((b,s), device=enc.device).masked_fill_(mel_pad_mask, 1)
stop_logits = self.final_norm(enc)
stop_logits = self.stop_head(stop_logits)
loss = F.binary_cross_entropy_with_logits(stop_logits.squeeze(-1), targets)
return loss.mean()
@register_model
def register_gpt_segmentor(opt_net, opt):
return GptSegmentor(**opt_get(opt_net, ['kwargs'], {}))
if __name__ == '__main__':
gpt = GptSegmentor()
l = gpt(torch.randn(3,80,94),
torch.tensor([18,42,93]))
print(l.shape)
#o = gpt.infer(torch.randint(high=24, size=(2,60)))
#print(o.shape)