generate-speaker-identification-apk-script.py
4.8 KB
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
#!/usr/bin/env python3
import argparse
from dataclasses import dataclass
from typing import List, Optional
import jinja2
def get_args():
parser = argparse.ArgumentParser()
parser.add_argument(
"--total",
type=int,
default=1,
help="Number of runners",
)
parser.add_argument(
"--index",
type=int,
default=0,
help="Index of the current runner",
)
return parser.parse_args()
@dataclass
class SpeakerIdentificationModel:
model_name: str
short_name: str = ""
lang: str = ""
framework: str = ""
def get_3dspeaker_models() -> List[SpeakerIdentificationModel]:
models = [
SpeakerIdentificationModel(model_name="3dspeaker_speech_campplus_sv_en_voxceleb_16k.onnx"),
SpeakerIdentificationModel(model_name="3dspeaker_speech_campplus_sv_zh-cn_16k-common.onnx"),
SpeakerIdentificationModel(model_name="3dspeaker_speech_eres2net_base_200k_sv_zh-cn_16k-common.onnx"),
SpeakerIdentificationModel(model_name="3dspeaker_speech_eres2net_base_sv_zh-cn_3dspeaker_16k.onnx"),
SpeakerIdentificationModel(model_name="3dspeaker_speech_eres2net_large_sv_zh-cn_3dspeaker_16k.onnx"),
SpeakerIdentificationModel(model_name="3dspeaker_speech_eres2net_sv_en_voxceleb_16k.onnx"),
SpeakerIdentificationModel(model_name="3dspeaker_speech_eres2net_sv_zh-cn_16k-common.onnx"),
]
prefix = '3dspeaker_speech_'
num = len(prefix)
for m in models:
m.framework = '3dspeaker'
m.short_name = m.model_name[num:-5]
if '_zh-cn_' in m.model_name:
m.lang = 'zh'
elif '_en_' in m.model_name:
m.lang = 'en'
else:
raise ValueError(m)
return models
def get_wespeaker_models() -> List[SpeakerIdentificationModel]:
models = [
SpeakerIdentificationModel(model_name="wespeaker_en_voxceleb_CAM++.onnx"),
SpeakerIdentificationModel(model_name="wespeaker_en_voxceleb_CAM++_LM.onnx"),
SpeakerIdentificationModel(model_name="wespeaker_en_voxceleb_resnet152_LM.onnx"),
SpeakerIdentificationModel(model_name="wespeaker_en_voxceleb_resnet221_LM.onnx"),
SpeakerIdentificationModel(model_name="wespeaker_en_voxceleb_resnet293_LM.onnx"),
SpeakerIdentificationModel(model_name="wespeaker_en_voxceleb_resnet34.onnx"),
SpeakerIdentificationModel(model_name="wespeaker_en_voxceleb_resnet34_LM.onnx"),
SpeakerIdentificationModel(model_name="wespeaker_zh_cnceleb_resnet34.onnx"),
SpeakerIdentificationModel(model_name="wespeaker_zh_cnceleb_resnet34_LM.onnx"),
]
prefix = 'wespeaker_xx_'
num = len(prefix)
for m in models:
m.framework = 'wespeaker'
m.short_name = m.model_name[num:-5]
if '_zh_' in m.model_name:
m.lang = 'zh'
elif '_en_' in m.model_name:
m.lang = 'en'
else:
raise ValueError(m)
return models
def get_nemo_models() -> List[SpeakerIdentificationModel]:
models = [
SpeakerIdentificationModel(model_name="nemo_en_speakerverification_speakernet.onnx"),
SpeakerIdentificationModel(model_name="nemo_en_titanet_large.onnx"),
SpeakerIdentificationModel(model_name="nemo_en_titanet_small.onnx"),
]
prefix = 'nemo_en_'
num = len(prefix)
for m in models:
m.framework = 'nemo'
m.short_name = m.model_name[num:-5]
if '_zh_' in m.model_name:
m.lang = 'zh'
elif '_en_' in m.model_name:
m.lang = 'en'
else:
raise ValueError(m)
return models
def main():
args = get_args()
index = args.index
total = args.total
assert 0 <= index < total, (index, total)
all_model_list = get_3dspeaker_models()
all_model_list += get_wespeaker_models()
all_model_list += get_nemo_models()
num_models = len(all_model_list)
num_per_runner = num_models // total
if num_per_runner <= 0:
raise ValueError(f"num_models: {num_models}, num_runners: {total}")
start = index * num_per_runner
end = start + num_per_runner
remaining = num_models - args.total * num_per_runner
print(f"{index}/{total}: {start}-{end}/{num_models}")
d = dict()
d["model_list"] = all_model_list[start:end]
if index < remaining:
s = args.total * num_per_runner + index
d["model_list"].append(all_model_list[s])
print(f"{s}/{num_models}")
filename_list = ["./build-apk-speaker-identification.sh"]
for filename in filename_list:
environment = jinja2.Environment()
with open(f"{filename}.in") as f:
s = f.read()
template = environment.from_string(s)
s = template.render(**d)
with open(filename, "w") as f:
print(s, file=f)
if __name__ == "__main__":
main()