File size: 2,057 Bytes
107eeac
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
import sys

import torch
from transformers import DebertaV2Model, DebertaV2Tokenizer

from config import config


LOCAL_PATH = "./bert/deberta-v3-large"

tokenizer = DebertaV2Tokenizer.from_pretrained(LOCAL_PATH)

models = dict()


def get_bert_feature(
    text,
    word2ph,
    device=config.bert_gen_config.device,
    assist_text=None,
    assist_text_weight=0.7,
):
    if (
        sys.platform == "darwin"
        and torch.backends.mps.is_available()
        and device == "cpu"
    ):
        device = "mps"
    if not device:
        device = "cuda"
    if device == "cuda" and not torch.cuda.is_available():
        device = "cpu"
    if device not in models.keys():
        models[device] = DebertaV2Model.from_pretrained(LOCAL_PATH).to(device)
    with torch.no_grad():
        inputs = tokenizer(text, return_tensors="pt")
        for i in inputs:
            inputs[i] = inputs[i].to(device)
        res = models[device](**inputs, output_hidden_states=True)
        res = torch.cat(res["hidden_states"][-3:-2], -1)[0].cpu()
        if assist_text:
            style_inputs = tokenizer(assist_text, return_tensors="pt")
            for i in style_inputs:
                style_inputs[i] = style_inputs[i].to(device)
            style_res = models[device](**style_inputs, output_hidden_states=True)
            style_res = torch.cat(style_res["hidden_states"][-3:-2], -1)[0].cpu()
            style_res_mean = style_res.mean(0)
    assert len(word2ph) == res.shape[0], (text, res.shape[0], len(word2ph))
    word2phone = word2ph
    phone_level_feature = []
    for i in range(len(word2phone)):
        if assist_text:
            repeat_feature = (
                res[i].repeat(word2phone[i], 1) * (1 - assist_text_weight)
                + style_res_mean.repeat(word2phone[i], 1) * assist_text_weight
            )
        else:
            repeat_feature = res[i].repeat(word2phone[i], 1)
        phone_level_feature.append(repeat_feature)

    phone_level_feature = torch.cat(phone_level_feature, dim=0)

    return phone_level_feature.T