blob: a15a29f9d5745911d969690ff427da272c2cd50e (
plain) (
blame)
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
|
{ lib
, python3
, fetchFromGitHub
, espeak-ng
}:
# USAGE:
# $ tts-server --list_models
# # pick your favorite vocoder/tts model
# $ tts-server --model_name tts_models/en/ljspeech/glow-tts --vocoder_name vocoder_models/universal/libri-tts/fullband-melgan
#
# If you upgrade from an old version you may have to delete old models from ~/.local/share/tts
#
# For now, for deployment check the systemd unit in the pull request:
# https://github.com/NixOS/nixpkgs/pull/103851#issue-521121136
let
python = python3.override {
packageOverrides = self: super: {
# API breakage with 0.9.0
# TypeError: mel() takes 0 positional arguments but 2 positional arguments (and 3 keyword-only arguments) were given
librosa = super.librosa.overridePythonAttrs (oldAttrs: rec {
version = "0.8.1";
src = oldAttrs.src.override {
inherit version;
sha256 = "c53d05e768ae4a3e553ae21c2e5015293e5efbfd5c12d497f1104cb519cca6b3";
};
});
};
};
in
python.pkgs.buildPythonApplication rec {
pname = "tts";
version = "0.6.1";
format = "setuptools";
src = fetchFromGitHub {
owner = "coqui-ai";
repo = "TTS";
rev = "v${version}";
sha256 = "sha256-YzMR/Tl1UvjdSqV/h4lYR6DuarEqEIM7RReqYznFU4Q=";
};
postPatch = let
relaxedConstraints = [
"gruut"
"librosa"
"mecab-python3"
"numba"
"numpy"
"umap-learn"
"unidic-lite"
];
in ''
sed -r -i \
${lib.concatStringsSep "\n" (map (package:
''-e 's/${package}.*[<>=]+.*/${package}/g' \''
) relaxedConstraints)}
requirements.txt
sed -i '/tensorboardX/d' requirements.txt
'';
nativeBuildInputs = with python.pkgs; [
cython
];
propagatedBuildInputs = with python.pkgs; [
anyascii
coqpit
coqui-trainer
flask
fsspec
gdown
gruut
inflect
jieba
librosa
matplotlib
mecab-python3
numba
pandas
pypinyin
pysbd
pytorch-bin
pyworld
scipy
soundfile
tensorflow
torchaudio-bin
tqdm
umap-learn
unidic-lite
webrtcvad
];
postInstall = ''
cp -r TTS/server/templates/ $out/${python.sitePackages}/TTS/server
# cython modules are not installed for some reasons
(
cd TTS/tts/utils/monotonic_align
${python.interpreter} setup.py install --prefix=$out
)
'';
checkInputs = with python.pkgs; [
espeak-ng
pytestCheckHook
];
preCheck = ''
# use the installed TTS in $PYTHONPATH instead of the one from source to also have cython modules.
mv TTS{,.old}
export PATH=$out/bin:$PATH
# numba tries to write to HOME directory
export HOME=$TMPDIR
for file in $(grep -rl 'python TTS/bin' tests); do
substituteInPlace "$file" \
--replace "python TTS/bin" "${python.interpreter} $out/lib/${python.libPrefix}/site-packages/TTS/bin"
done
'';
disabledTests = [
# Requires network acccess to download models
"test_synthesize"
"test_run_all_models"
# Mismatch between phonemes
"test_text_to_ids_phonemes_with_eos_bos_and_blank"
# Takes too long
"test_parametrized_wavernn_dataset"
];
disabledTestPaths = [
# phonemes mismatch between espeak-ng and gruuts phonemizer
"tests/text_tests/test_phonemizer.py"
# no training, it takes too long
"tests/aux_tests/test_speaker_encoder_train.py"
"tests/tts_tests/test_align_tts_train.py"
"tests/tts_tests/test_fast_pitch_speaker_emb_train.py"
"tests/tts_tests/test_fast_pitch_train.py"
"tests/tts_tests/test_glow_tts_d-vectors_train.py"
"tests/tts_tests/test_glow_tts_speaker_emb_train.py"
"tests/tts_tests/test_glow_tts_train.py"
"tests/tts_tests/test_speedy_speech_train.py"
"tests/tts_tests/test_tacotron2_d-vectors_train.py"
"tests/tts_tests/test_tacotron2_speaker_emb_train.py"
"tests/tts_tests/test_tacotron2_train.py"
"tests/tts_tests/test_tacotron2_train_fsspec_path.py"
"tests/tts_tests/test_tacotron_train.py"
"tests/tts_tests/test_vits_d-vectors_train.py"
"tests/tts_tests/test_vits_multilingual_speaker_emb_train.py"
"tests/tts_tests/test_vits_multilingual_train-d_vectors.py"
"tests/tts_tests/test_vits_speaker_emb_train.py"
"tests/tts_tests/test_vits_train.py"
"tests/vocoder_tests/test_wavegrad_train.py"
"tests/vocoder_tests/test_parallel_wavegan_train.py"
"tests/vocoder_tests/test_fullband_melgan_train.py"
"tests/vocoder_tests/test_hifigan_train.py"
"tests/vocoder_tests/test_multiband_melgan_train.py"
"tests/vocoder_tests/test_melgan_train.py"
"tests/vocoder_tests/test_wavernn_train.py"
];
meta = with lib; {
homepage = "https://github.com/coqui-ai/TTS";
changelog = "https://github.com/coqui-ai/TTS/releases/tag/v${version}";
description = "Deep learning toolkit for Text-to-Speech, battle-tested in research and production";
license = licenses.mpl20;
maintainers = teams.tts.members;
};
}
|