AudioGPT/NeuralSeq/tasks/tts/__pycache__/fs2_utils.cpython-38.pyc

U
<00><>dS<00>@s<>dd
lZe<00>
d<02>

dd
lZdd
lZddlmZ
dd
lZdd
lZdd
l	Zddl
mZ
ddlm
Z

dd
lZddlmZ
dd
lZdd
lZdd
l	Zdd
lZdd
lZddlmZ
Gdd	<09>d	e<11>Zd
S)
<EFBFBD>N<>Agg)
<01>get_lf0_cwt)
<01>IndexedDataset)
<01>norm_interp_f0)
<01>BaseDataset)
<01>hparamscs@eZ
dZd
<0A>f
dd<03>	Zdd<05>Zdd<07>Zdd	<09>Zddd<0C>
Z<07>ZS)<0F>FastSpeechDatasetFcsP
t<00><00>
|<02>

td
<00>_|
<01>_t<02>_t<05><06>j<03>d<02>j<04>d<03><04>
<01>_d<00>_<08>j<03>d<04>}t	j
<EFBFBD>|<03>
r<>t<05>|<03>
\td<td<\<02>_<0C>_
ttd<00>
td<ttd<00>
td<nd\td<td<\<02>_<0C>_
|
dk<02>
r(td	d
kr<><72><00>td	<00>
\<02>_<08>_n>tddk<04>
r(tttd<00>
<01>
td
<00>_<12>f
dd<0F><08>jD<00>
<01>_tddk<02>
rLtt<05>d<12>
<01>
\}td<dS)N<>binary_data_dir<69>
/z_lengths.npyz/train_f0s_mean_std.npy<70>f0_mean<61>f0_std)NN<4E>test<73>test_input_dir<69>Znum_test_samplesr
Ztest_idsc
sg|]}
<01>j|
<00>qS<00>)
<01>sizes)<02>.0<EFBFBD>
i<>
<01>selfr<00>S/mnt/sdc/hongzhiqing/github/AudioGPT/text_to_sing/DiffSinger/tasks/tts/fs2_utils.py<70>
<listcomp>0sz.FastSpeechDataset.__init__.<locals>.<listcomp><3E>
pitch_type<EFBFBD>cwt<77>
<00>
cwt_scales)<15>super<65>__init__r<00>data_dir<69>prefix<69>np<6E>loadr<00>
indexed_ds<EFBFBD>os<6F>path<74>existsrr<00>float<61>load_test_inputs<74>list<73>range<67>
avail_idxsr<00>ones)rr<00>shuffle<6C>f0_stats_fn<66>
_<>
<01>	__class__rrrs(








"







zFastSpeechDataset.__init__cCsJt|d
<01>r|j
dk	r|j
|
}
|jdkr@t|j<04>d|j<05><00><03>
|_|j|
S)Nr*r
)<06>hasattrr*r"rrr)r<00>indexrrr<00>	_get_item5s







zFastSpeechDataset._get_itemc
Cs<>
|j}|<00>
|
<01>
}|d
}t<02>|d<00>
d|<04>}|<05><04>d<00>d<04>
<01><06>}d|krbt<02>|d<00>
d|<04>nd}t|dd|<04>|<02>\}}	t<02>|dd|d<00><00>
}
t<02>|<03>	d	<09>
<01>
d|<04>}|
|d
|d|
|||||	||<05>
<EFBFBD><00>d<04>
dkd
<0A>}|jd<00>
rt<02>|d<00>
|d<|jd<00>
r|d|d<|jddk<02>
rxt<02>|d<00>
d|<04>}
|<03>	d|<03>	d<16>
<01>}|<03>	d|<03>	d<18>
<01>}|<0C>|
||d<19><03>

n`|jddk<02>
r<>t<02>|
<EFBFBD>
<01>
<0A><00>d|d|<08>}t<02>|
<EFBFBD>
<01>
<0A><00>d|dt<02>|<08>
<01><03>d<1B>
}|||d<|S)N<>
max_frames<EFBFBD>mel<65><00><><EFBFBD><EFBFBD><EFBFBD><EFBFBD>mel2ph<70>f0<66>phoneZmax_input_tokens<6E>pitch<63>	item_name<6D>txtr
)<0B>idr<<00>text<78>	txt_tokenr5r;<00>energyr9<00>uvr8Zmel_nonpadding<6E>
use_spk_embed<65>	spk_embed<65>
use_spk_id<EFBFBD>spk_idrr<00>cwt_specrZcwt_meanrZcwt_std<74>rGrr<00>ph<70>
<00>f0_ph)rr3<00>torch<63>Tensor<6F>exp<78>sum<75>sqrt<72>
LongTensorr<00>get<65>abs<62>update<74>
zeros_liker&<00>scatter_add<64>	ones_like<6B>	clamp_min)rr2r<00>itemr4<00>specrAr8r9rBr:r;<00>samplerGrrZf0_phlevel_sumZf0_phlevel_numrrr<00>__getitem__<sR






"












<10>











<08>
<02>
zFastSpeechDataset.__getitem__cCst|
<01>
d
kriSt
<01>dd<03>|
D<00>
<01>
}dd<03>|
D<00>
}dd<03>|
D<00>
}t<03>dd<03>|
D<00>
d
<01>}t<03>dd<03>|
D<00>
d<08>}t<03>d	d<03>|
D<00>
<01>
}t<03>d
d<03>|
D<00>
<01>
}t<03>dd<03>|
D<00>
d<08>}	|
d
ddk	r<>t<03>d
d<03>|
D<00>
d<08>nd}
t<03>dd<03>|
D<00>
d<08>}t
<01>dd<03>|
D<00>
<01>
}t
<01>dd<03>|
D<00>
<01>
}
||t|
<01>
|||||
|
|	|||d<11>
}|jd<00>
r^t
<01>dd<03>|
D<00>
<01>
}||d<|jd<00>
r<>t
<01>dd<03>|
D<00>
<01>
}||d<|jddk<02>
r<>t<03>dd<03>|
D<00>
<01>
}t
<01>dd<03>|
D<00>
<01>
}t
<01>dd<03>|
D<00>
<01>
}|<0E>	|||d<1D><03>

n(|jddk<02>rt<03>dd<03>|
D<00>
<01>
|d <|S)!Nr
c
Ssg|]}
|
d<00>qS)
r>r<00>r<00>
srrrrlsz.FastSpeechDataset.collater.<locals>.<listcomp>c
Ssg|]}
|
d<00>qS)
r<rr]rrrrmsc
Ssg|]}
|
d<00>qS)
r?rr]rrrrnsc
Ssg|]}
|
d<00>qS<00>
r@rr]rrrrosc
Ssg|]}
|
d<00>qS)
r9rr]rrrrpsgc
Ssg|]}
|
d<00>qS)
r;rr]rrrrqsc
Ssg|]}
|
d<00>qS)
rBrr]rrrrrsc
Ssg|]}
|
d<00>qS)
rArr]rrrrssr8c
Ssg|]}
|
d<00>qS)
r8rr]rrrrtsc
Ssg|]}
|
d<00>qS)
r5rr]rrrrvsc
Ssg|]}
|
d<00><00><00>qSr_)
<01>numelr]rrrrwsc
Ssg|]}
|
djd
<00>qS)r5r
)
<01>shaper]rrrrxs)
r>r<<00>nsamplesr?<00>
txt_tokens<EFBFBD>txt_lengths<68>mels<6C>mel_lengthsr8rAr;r9rBrCc
Ssg|]}
|
d<00>qS)
rDrr]rrrr<00>srDrEc
Ssg|]}
|
d<00>qS)
rFrr]rrrr<00>s<00>spk_idsrrc
Ssg|]}
|
d<00>qS)
rGrr]rrrr<00>sc
Ssg|]}
|
d<00>qS)
rrr]rrrr<00>sc
Ssg|]}
|
d<00>qS)
rrr]rrrr<00>srHrIc
Ssg|]}
|
d<00>qS)
rKrr]rrrr<00>sr9)
<EFBFBD>lenrLrQ<00>utils<6C>
collate_1d<EFBFBD>
collate_2dr<00>stackrMrT)r<00>samplesr><00>
item_namesr?rcr9r;rBrAr8rerdrf<00>batchrDrgrGrrrrr<00>collaterisZ









<0E>
<02>













<02>











zFastSpeechDataset.collaterr
cCs<>t<00>|
<01>d
<01><02>
t<00>|
<01>d<02><02>
}g}g}t
<01>dd<04>}d<05>|<06>d<05>
dd<06><00>
}|<06>d<05>
d}tt<06>|<07>
|<08>}t
d}	|D]T}
tj	<09>
|
<EFBFBD>
}d}}
}|
}
d}|<06>|||
||
|||	<09>}|<05>|<10>

|<04>|d	<00>

qv||fS)
Nz/*.wavz/*.mp3<70>
binarizer_clsz*data_gen.tts.base_binarizerr.BaseBinarizer<65>
.r7<00>binarization_argsrrh)
<0A>globrrR<00>join<69>split<69>getattr<74>	importlib<69>
import_moduler#r$<00>basenameZprocess_item<65>append)rrrFZ
inp_wav_pathsr<00>itemsrq<00>pkg<6B>cls_namers<00>wav_fnr<rIr=<00>tg_fn<66>encoderrYrrrr'<00>s"
 














z"FastSpeechDataset.load_test_inputs)
F)
r
)	<09>__name__<5F>
__module__<EFBFBD>__qualname__rr3r\rpr'<00>
__classcell__rrr/rrs

-1r)<17>
matplotlib<EFBFBD>usertrx<00>	utils.cwtrr#<00>torch.optimrL<00>torch.utils.dataZutils.indexed_datasetsr<00>utils.pitch_utilsr<00>numpyr Ztasks.base_taskrri<00>torch.distributions<6E>
utils.hparamsrrrrrr<00><module>
s$