nixpkgs/pkgs/tools/audio/tts/default.nix
2020-11-15 04:01:15 +01:00

124 lines
3.3 KiB
Nix

{ lib
, python3Packages
, fetchFromGitHub
, fetchpatch
, python3
}:
#
# Tested in the following setup:
#
# TTS model:
# Tacotron2 DDC
# https://drive.google.com/drive/folders/1Y_0PcB7W6apQChXtbt6v3fAiNwVf4ER5
# Vocoder model:
# Multi-Band MelGAN
# https://drive.google.com/drive/folders/1XeRT0q4zm5gjERJqwmX5w84pMrD00cKD
#
# Arrange /tmp/tts like this:
# scale_stats.npy
# tts
# tts/checkpoint_130000.pth.tar
# tts/checkpoint_130000_tf.pkl
# tts/checkpoint_130000_tf_2.3rc0.tflite
# tts/config.json
# tts/scale_stats.npy
# vocoder
# vocoder/checkpoint_1450000.pth.tar
# vocoder/checkpoint_2750000_tf.pkl
# vocoder/checkpoint_2750000_tf_v2.3rc.tflite
# vocoder/config.json
# vocoder/scale_stats.npy
#
# Start like this:
# cd /tmp/tts
# tts-server \
# --vocoder_config ./tts/vocoder/config.json \
# --vocoder_checkpoint ./tts/vocoder/checkpoint_1450000.pth.tar \
# --tts_config ./tts/config.json \
# --tts_checkpoint ./tts/checkpoint_130000.pth.tar
#
# For now, for deployment check the systemd unit in the pull request:
# https://github.com/NixOS/nixpkgs/pull/103851#issue-521121136
#
python3Packages.buildPythonApplication rec {
pname = "tts";
# until https://github.com/mozilla/TTS/issues/424 is resolved
# we treat released models as released versions:
# https://github.com/mozilla/TTS/wiki/Released-Models
version = "unstable-2020-06-17";
src = fetchFromGitHub {
owner = "mozilla";
repo = "TTS";
rev = "72a6ac54c8cfaa407fc64b660248c6a788bdd381";
sha256 = "1wvs264if9n5xzwi7ryxvwj1j513szp6sfj6n587xk1fphi0921f";
};
patches = [
(fetchpatch {
url = "https://github.com/mozilla/TTS/commit/36fee428b9f3f4ec1914b090a2ec9d785314d9aa.patch";
sha256 = "sha256-pP0NxiyrsvQ0A7GEleTdT87XO08o7WxPEpb6Bmj66dc=";
})
];
preBuild = ''
# numba jit tries to write to its cache directory
export HOME=$TMPDIR
sed -i -e 's!tensorflow==.*!tensorflow!' requirements.txt
sed -i -e 's!librosa==[^"]*!librosa!' requirements.txt setup.py
sed -i -e 's!unidecode==[^"]*!unidecode!' requirements.txt setup.py
sed -i -e 's!bokeh==[^"]*!bokeh!' requirements.txt setup.py
sed -i -e 's!numba==[^"]*!numba!' requirements.txt setup.py
# Not required for building/installation but for their development/ci workflow
sed -i -e '/pylint/d' requirements.txt setup.py
sed -i -e '/cardboardlint/d' requirements.txt setup.py
'';
propagatedBuildInputs = with python3Packages; [
matplotlib
scipy
pytorch
flask
attrdict
bokeh
soundfile
tqdm
librosa
unidecode
phonemizer
tensorboardx
fuzzywuzzy
tensorflow_2
inflect
gdown
pysbd
];
postInstall = ''
cp -r TTS/server/templates/ $out/${python3.sitePackages}/TTS/server
'';
checkInputs = with python3Packages; [ pytestCheckHook ];
disabledTests = [
# RuntimeError: fft: ATen not compiled with MKL support
"test_torch_stft"
"test_stft_loss"
"test_multiscale_stft_loss"
# AssertionErrors that I feel incapable of debugging
"test_phoneme_to_sequence"
"test_text2phone"
"test_parametrized_gan_dataset"
];
meta = with lib; {
homepage = "https://github.com/mozilla/TTS";
description = "Deep learning for Text to Speech";
license = licenses.mpl20;
maintainers = with maintainers; [ hexa mic92 ];
};
}