File size: 7,013 Bytes
7e9eac8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ebb01fc
7e9eac8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ebb01fc
7e9eac8
 
0969649
7e9eac8
 
 
 
 
 
 
40de727
4cd7d34
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
40de727
588b97f
f7215c7
0ea07f8
684c796
c5bf357
9462bda
c5bf357
 
9462bda
 
 
 
 
4cd7d34
684c796
0ea07f8
 
4cd7d34
 
bd7955f
4cd7d34
 
 
 
 
 
 
 
 
 
 
 
 
dd6a80b
0ea07f8
4cd7d34
 
 
 
 
 
 
 
 
 
 
 
7e9eac8
684c796
 
 
 
 
7e9eac8
ebb01fc
7e9eac8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
23e7dca
 
 
 
 
 
7e9eac8
 
 
 
 
 
ebb01fc
7e9eac8
 
0969649
7e9eac8
23e7dca
 
7e9eac8
 
 
 
 
 
d5f48a5
51a3cc9
407c586
51a3cc9
 
 
d5f48a5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c5bf357
 
 
 
 
d5f48a5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7e9eac8
ebb01fc
cd6ff39
9c1cd2f
7e9eac8
 
 
9c1cd2f
d5f48a5
9e48595
 
7e9eac8
9c1cd2f
7e9eac8
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
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
# Copyright      2022-2023  Xiaomi Corp.        (authors: Fangjun Kuang)
#
# See LICENSE for clarification regarding multiple authors
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

from functools import lru_cache

import sherpa_onnx
from huggingface_hub import hf_hub_download


def get_file(
    repo_id: str,
    filename: str,
    subfolder: str = ".",
) -> str:
    model_filename = hf_hub_download(
        repo_id=repo_id,
        filename=filename,
        subfolder=subfolder,
    )
    return model_filename


@lru_cache(maxsize=10)
def _get_vits_vctk(repo_id: str, speed: float) -> sherpa_onnx.OfflineTts:
    assert repo_id == "csukuangfj/vits-vctk"

    model = get_file(
        repo_id=repo_id,
        filename="vits-vctk.onnx",
        subfolder=".",
    )

    lexicon = get_file(
        repo_id=repo_id,
        filename="lexicon.txt",
        subfolder=".",
    )

    tokens = get_file(
        repo_id=repo_id,
        filename="tokens.txt",
        subfolder=".",
    )

    tts_config = sherpa_onnx.OfflineTtsConfig(
        model=sherpa_onnx.OfflineTtsModelConfig(
            vits=sherpa_onnx.OfflineTtsVitsModelConfig(
                model=model,
                lexicon=lexicon,
                tokens=tokens,
                length_scale=1.0 / speed,
            ),
            provider="cpu",
            debug=True,
            num_threads=2,
        )
    )
    tts = sherpa_onnx.OfflineTts(tts_config)

    return tts


@lru_cache(maxsize=10)
def _get_vits_ljs(repo_id: str, speed: float) -> sherpa_onnx.OfflineTts:
    assert repo_id == "csukuangfj/vits-ljs"

    model = get_file(
        repo_id=repo_id,
        filename="vits-ljs.onnx",
        subfolder=".",
    )

    lexicon = get_file(
        repo_id=repo_id,
        filename="lexicon.txt",
        subfolder=".",
    )

    tokens = get_file(
        repo_id=repo_id,
        filename="tokens.txt",
        subfolder=".",
    )

    tts_config = sherpa_onnx.OfflineTtsConfig(
        model=sherpa_onnx.OfflineTtsModelConfig(
            vits=sherpa_onnx.OfflineTtsVitsModelConfig(
                model=model,
                lexicon=lexicon,
                tokens=tokens,
                length_scale=1.0 / speed,
            ),
            provider="cpu",
            debug=True,
            num_threads=2,
        )
    )
    tts = sherpa_onnx.OfflineTts(tts_config)

    return tts


@lru_cache(maxsize=10)
def _get_vits_piper(repo_id: str, speed: float) -> sherpa_onnx.OfflineTts:
    data_dir = "/tmp/espeak-ng-data"
    if "coqui" in repo_id or "vits-mms" in repo_id:
        name = "model"
    elif "piper" in repo_id:
        n = len("vits-piper-")
        name = repo_id.split("/")[1][n:]
    elif "mimic3" in repo_id:
        n = len("vits-mimic3-")
        name = repo_id.split("/")[1][n:]
    else:
        raise ValueError(f"Unsupported {repo_id}")

    if "vits-coqui-uk-mai" in repo_id or "vits-mms" in repo_id:
        data_dir = ""

    model = get_file(
        repo_id=repo_id,
        filename=f"{name}.onnx",
        subfolder=".",
    )

    tokens = get_file(
        repo_id=repo_id,
        filename="tokens.txt",
        subfolder=".",
    )

    tts_config = sherpa_onnx.OfflineTtsConfig(
        model=sherpa_onnx.OfflineTtsModelConfig(
            vits=sherpa_onnx.OfflineTtsVitsModelConfig(
                model=model,
                lexicon="",
                data_dir=data_dir,
                tokens=tokens,
                length_scale=1.0 / speed,
            ),
            provider="cpu",
            debug=True,
            num_threads=2,
        )
    )
    tts = sherpa_onnx.OfflineTts(tts_config)

    return tts


@lru_cache(maxsize=10)
def _get_vits_mms(repo_id: str, speed: float) -> sherpa_onnx.OfflineTts:
    return _get_vits_piper(repo_id, speed)


@lru_cache(maxsize=10)
def _get_vits_zh_aishell3(repo_id: str, speed: float) -> sherpa_onnx.OfflineTts:
    assert repo_id == "csukuangfj/vits-zh-aishell3"

    model = get_file(
        repo_id=repo_id,
        filename="vits-aishell3.onnx",
        subfolder=".",
    )

    lexicon = get_file(
        repo_id=repo_id,
        filename="lexicon.txt",
        subfolder=".",
    )

    tokens = get_file(
        repo_id=repo_id,
        filename="tokens.txt",
        subfolder=".",
    )

    rule_fst = get_file(
        repo_id=repo_id,
        filename="rule.fst",
        subfolder=".",
    )

    tts_config = sherpa_onnx.OfflineTtsConfig(
        model=sherpa_onnx.OfflineTtsModelConfig(
            vits=sherpa_onnx.OfflineTtsVitsModelConfig(
                model=model,
                lexicon=lexicon,
                tokens=tokens,
                length_scale=1.0 / speed,
            ),
            provider="cpu",
            debug=True,
            num_threads=2,
        ),
        rule_fsts=rule_fst,
    )
    tts = sherpa_onnx.OfflineTts(tts_config)

    return tts


@lru_cache(maxsize=10)
def _get_vits_hf(repo_id: str, speed: float) -> sherpa_onnx.OfflineTts:
    if "fanchen" in repo_id or "vits-cantonese-hf-xiaomaiiwn" in repo_id:
        model = repo_id.split("/")[-1]
    else:
        model = repo_id.split("-")[-1]

    model = get_file(
        repo_id=repo_id,
        filename=f"{model}.onnx",
        subfolder=".",
    )

    lexicon = get_file(
        repo_id=repo_id,
        filename="lexicon.txt",
        subfolder=".",
    )

    tokens = get_file(
        repo_id=repo_id,
        filename="tokens.txt",
        subfolder=".",
    )

    rule_fst = get_file(
        repo_id=repo_id,
        filename="rule.fst",
        subfolder=".",
    )

    tts_config = sherpa_onnx.OfflineTtsConfig(
        model=sherpa_onnx.OfflineTtsModelConfig(
            vits=sherpa_onnx.OfflineTtsVitsModelConfig(
                model=model,
                lexicon=lexicon,
                tokens=tokens,
                length_scale=1.0 / speed,
            ),
            provider="cpu",
            debug=True,
            num_threads=2,
        ),
        rule_fsts=rule_fst,
    )
    tts = sherpa_onnx.OfflineTts(tts_config)

    return tts


@lru_cache(maxsize=10)
def get_pretrained_model(repo_id: str, speed: float) -> sherpa_onnx.OfflineTts:
    if repo_id in spanish_models:
        return spanish_models[repo_id](repo_id, speed)
    else:
        raise ValueError(f"Unsupported repo_id: {repo_id}")

spanish_models = {
    "csukuangfj/vits-piper-es_ES-sharvard-medium": _get_vits_piper,  # 2 speakers
}

language_to_models = {
    "Spanish": list(spanish_models.keys()),
}