h< SSKrSSKJr SSKJrJrJrJrJrJ r SSK r SSK J r SSK J r SSKJr SSKJrJr SSKJrJr S S KJr S S KJr /rS r"S S\R85r"SS\R85r"SS\ R>R@\RB5r""SS\ R>R@\RB5r#"SS5r$"SS5r%\"SS55r&\"SS55r'"SS5r(\"SS \'\&\$\55r)\"S!S"\'\&\%\55r*\"S#S$\(\&\$\55r+\"S%S&\(\&\%\55r,\+"S'\RZ"S(S)9S*9r.S+\.l/\,"S,\RZ"S-S)9S*9r0S.\0l/\)"S/\RZ"S(S)9S0\Rb"5S19r2S2\2l/\*"S3\RZ"S-S)9S0\Rb"5S19r3S4\3l/g)5N) dataclass)AnyDictListOptionalTupleUnion)Tensor)load_state_dict_from_url)mu_law_decoding) Tacotron2WaveRNN) GriffinLimInverseMelScale)utils)Tacotron2TTSBundlez.https://download.pytorch.org/torchaudio/modelscd^\rSrSrU4Sjr\S5rS\\\ \4S\ \ \ 44Sjr Sr U=r$)_EnglishCharProcessorc>[TU]5 [R"5Ul[ UR5VVs0sHupX!_M snnUlgs snnfN)super__init__r _get_chars_tokens enumerate_mapping)selfis __class__s V/var/www/auris/envauris/lib/python3.13/site-packages/torchaudio/pipelines/_tts/impl.pyr_EnglishCharProcessor.__init__sF '') *3DLL*AB*A$!*AB BsAcUR$rrrs r#tokens_EnglishCharProcessor.tokens ||textsreturnc [U[5(aU/nUVVs/sHAo"R5Vs/sH#o3UR;dMURUPM% snPMC nnn[R "U5$s snfs snnfr) isinstancestrlowerrr _to_tensor)rr,tcindicess r#__call___EnglishCharProcessor.__call__#so eS ! !GEX]^X]STggiNi ;M$DMM!$iNX]^((O^sBA? A? B?B)rr__name__ __module__ __qualname____firstlineno__rpropertyr(r r0rrr r6__static_attributes__ __classcell__r"s@r#rrsKC )eCcN3)ffn8M))r+rcl^\rSrSrSS.U4Sjjr\S5rS\\\ \4S\ \ \ 44Sjr S r U=r$) _EnglishPhoneProcessor*N dl_kwargsc>[TU]5 [R"5Ul[ UR5VVs0sHup#X2_M snnUl[R"SUS9UlSUl gs snnf)Nzen_us_cmudict_forward.ptrDz(\[[A-Z]+?\]|[_!'(),.:;? -])) rrr _get_phonesrrr_load_phonemizer _phonemizer_pattern)rrEr pr"s r#r_EnglishPhoneProcessor.__init__+sg ((* *3DLL*AB*A$!*AB  112LXab7 CsA;cUR$rr&r's r#r(_EnglishPhoneProcessor.tokens2r*r+r,r-c [U[5(aU/n/nURUSS9Hxn[R"UR U5Vs/sHn[R "SSU5PM nnURUVs/sHo`RUPM sn5 Mz [R"U5$s snfs snf)Nen_us)langz[\[\]]) r/r0rIrefindallrJsubappendrrr2)rr,r5phonesrretrKs r#r6_EnglishPhoneProcessor.__call__6s eS ! !GE&&u7&;F57ZZ v5VW5V266)R+5VCW NNc:cMM!,c: ;<((X:s "B=C )rrJrIrr8r@s@r#rBrB*sR$(88 )eCcN3 )ffn8M ) )r+rBcX^\rSrSrSS\S\\4U4Sjjjr\S5r S Sjr Sr U=r $) _WaveRNNVocoderGmodel min_level_dbcH>[TU]5 SUlXlX lg)N"V)rr _sample_rate_model _min_level_db)rr^r_r"s r#r_WaveRNNVocoder.__init__Hs! ! )r+cUR$rrbr's r# sample_rate_WaveRNNVocoder.sample_rateN   r+c[R"U5nS[R"[R"USS95-nURb2URU- UR- n[R"USSS9nUR R X5up2[R"X0R R5n[X0R R5nURS5nX24$)Ngh㈵>)minrr)rmmax) torchexplog10clamprdrcinferr_unnormalize_waveformn_bitsr n_classessqueeze)rmel_speclengthswaveforms r#forward_WaveRNNVocoder.forwardRs99X& EKKd$CDD    )**X59K9KKH{{8:H KK--h@..x9K9KL"8[[-B-BC##A&  r+)rdrcrb)ir) r9r:r;r<rrfloatrr=rhr{r>r?r@s@r#r\r\Gs=*g*Xe_** !! ! !r+r\cB^\rSrSrU4Sjr\S5rSSjrSrU=r $)_GriffinLimVocoder_c >[TU]5 SUl[SSURSSSSS9Ul[ SS S SS 9Ulg) NraiPgg@@slaney)n_stftn_melsrhf_minf_max mel_scalenormir)n_fftpower hop_length win_length)rrrbrrh_inv_melr _griffin_lim)rr"s r#r_GriffinLimVocoder.__init__`sX !'!((  '  r+cUR$rrgr's r#rh_GriffinLimVocoder.sample_ratesrjr+c[R"U5nUR5R5R S5nUR U5nUR5R S5nUR U5nXB4$)NTF)rorpclonedetachrequires_grad_rr)rrxryspec waveformss r#r{_GriffinLimVocoder.forwardwsm99X&>>#**,;;DA}}X&{{}++E2%%d+ !!r+)rrrbr) r9r:r;r<rr=rhr{r>r?r@s@r#rr_s& &!!""r+rc6\rSrSrS\R 4SjrSrg) _CharMixinr-c[5$r)rr's r#get_text_processor_CharMixin.get_text_processors $&&r+Nr9r:r;r<r TextProcessorrr>rr+r#rrs'$6$D$D'r+rc>\rSrSrSS.S\R 4SjjrSrg) _PhoneMixinNrDr-c[US9$NrD)rB)rrEs r#r_PhoneMixin.get_text_processors % ::r+rrrr+r#rrs.2;7I7W7W;;r+rcJ\rSrSr%\\S'\\\4\S'SS.S\4Sjjr Sr g) _Tacotron2Mixin_tacotron2_path_tacotron2_paramsNrDr-c[S0URD6n[SUR3nUc0OUn[ U40UD6nUR U5 UR 5 U$N/r)r r _BASE_URLrr load_state_dictevalrrEr^url state_dicts r# get_tacotron2_Tacotron2Mixin.get_tacotron2sc3D223 1T1123#+B -c?Y?  j)  r+r) r9r:r;r<r0__annotations__rrr rr>rr+r#rrs*CH~%)-)r+rc\\rSrSr%\\\S'\\\\4\S'SS.Sjr SS.Sjr Sr g) _WaveRNNMixin _wavernn_path_wavernn_paramsNrDc6URUS9n[U5$r) _get_wavernnr\)rrEwavernns r# get_vocoder_WaveRNNMixin.get_vocoders ##i#8w''r+c[S0URD6n[SUR3nUc0OUn[ U40UD6nUR U5 UR 5 U$r)rrrrr rrrs r#r_WaveRNNMixin._get_wavernnsc/$../ 1T//01#+B -c?Y?  j)  r+r) r9r:r;r<rr0rrrrrr>rr+r#rrs6C= d38n--'+()-r+rc\rSrSrSrSrg)_GriffinLimMixinc [5$r)r)r_s r#r_GriffinLimMixin.get_vocoders !##r+rN)r9r:r;r<rr>rr+r#rrs$r+rc\rSrSrSrg)_Tacotron2WaveRNNCharBundlerNr9r:r;r<r>rr+r#rrr+rc\rSrSrSrg)_Tacotron2WaveRNNPhoneBundlerNrrr+r#rrrr+rc\rSrSrSrg)_Tacotron2GriffinLimCharBundlerNrrr+r#rrrr+rc\rSrSrSrg)_Tacotron2GriffinLimPhoneBundlerNrrr+r#rrrr+rz5tacotron2_english_characters_1500_epochs_ljspeech.pth&) n_symbols)rraCharacter-based TTS pipeline with :py:class:`~torchaudio.models.Tacotron2` trained on *LJSpeech* :cite:`ljspeech17` for 1,500 epochs, and :py:class:`~torchaudio.transforms.GriffinLim` as vocoder. The text processor encodes the input texts character-by-character. You can find the training script `here `__. The default parameters were used. Please refer to :func:`torchaudio.pipelines.Tacotron2TTSBundle` for the usage. Example - "Hello world! T T S stands for Text to Speech!" .. image:: https://download.pytorch.org/torchaudio/doc-assets/TACOTRON2_GRIFFINLIM_CHAR_LJSPEECH.png :alt: Spectrogram generated by Tacotron2 .. raw:: html Example - "The examination and testimony of the experts enabled the Commission to conclude that five shots may have been fired," .. image:: https://download.pytorch.org/torchaudio/doc-assets/TACOTRON2_GRIFFINLIM_CHAR_LJSPEECH_v2.png :alt: Spectrogram generated by Tacotron2 .. raw:: html z3tacotron2_english_phonemes_1500_epochs_ljspeech.pth`aPhoneme-based TTS pipeline with :py:class:`~torchaudio.models.Tacotron2` trained on *LJSpeech* :cite:`ljspeech17` for 1,500 epochs and :py:class:`~torchaudio.transforms.GriffinLim` as vocoder. The text processor encodes the input texts based on phoneme. It uses `DeepPhonemizer `__ to convert graphemes to phonemes. The model (*en_us_cmudict_forward*) was trained on `CMUDict `__. You can find the training script `here `__. The text processor is set to the *"english_phonemes"*. Please refer to :func:`torchaudio.pipelines.Tacotron2TTSBundle` for the usage. Example - "Hello world! T T S stands for Text to Speech!" .. image:: https://download.pytorch.org/torchaudio/doc-assets/TACOTRON2_GRIFFINLIM_PHONE_LJSPEECH.png :alt: Spectrogram generated by Tacotron2 .. raw:: html Example - "The examination and testimony of the experts enabled the Commission to conclude that five shots may have been fired," .. image:: https://download.pytorch.org/torchaudio/doc-assets/TACOTRON2_GRIFFINLIM_PHONE_LJSPEECH_v2.png :alt: Spectrogram generated by Tacotron2 .. raw:: html z=tacotron2_english_characters_1500_epochs_wavernn_ljspeech.pthz%wavernn_10k_epochs_8bits_ljspeech.pth)rrrraCharacter-based TTS pipeline with :py:class:`~torchaudio.models.Tacotron2` trained on *LJSpeech* :cite:`ljspeech17` for 1,500 epochs and :py:class:`~torchaudio.models.WaveRNN` vocoder trained on 8 bits depth waveform of *LJSpeech* :cite:`ljspeech17` for 10,000 epochs. The text processor encodes the input texts character-by-character. You can find the training script `here `__. The following parameters were used; ``win_length=1100``, ``hop_length=275``, ``n_fft=2048``, ``mel_fmin=40``, and ``mel_fmax=11025``. You can find the training script `here `__. Please refer to :func:`torchaudio.pipelines.Tacotron2TTSBundle` for the usage. Example - "Hello world! T T S stands for Text to Speech!" .. image:: https://download.pytorch.org/torchaudio/doc-assets/TACOTRON2_WAVERNN_CHAR_LJSPEECH.png :alt: Spectrogram generated by Tacotron2 .. raw:: html Example - "The examination and testimony of the experts enabled the Commission to conclude that five shots may have been fired," .. image:: https://download.pytorch.org/torchaudio/doc-assets/TACOTRON2_WAVERNN_CHAR_LJSPEECH_v2.png :alt: Spectrogram generated by Tacotron2 .. raw:: html z;tacotron2_english_phonemes_1500_epochs_wavernn_ljspeech.pthaPhoneme-based TTS pipeline with :py:class:`~torchaudio.models.Tacotron2` trained on *LJSpeech* :cite:`ljspeech17` for 1,500 epochs, and :py:class:`~torchaudio.models.WaveRNN` vocoder trained on 8 bits depth waveform of *LJSpeech* :cite:`ljspeech17` for 10,000 epochs. The text processor encodes the input texts based on phoneme. It uses `DeepPhonemizer `__ to convert graphemes to phonemes. The model (*en_us_cmudict_forward*) was trained on `CMUDict `__. You can find the training script for Tacotron2 `here `__. The following parameters were used; ``win_length=1100``, ``hop_length=275``, ``n_fft=2048``, ``mel_fmin=40``, and ``mel_fmax=11025``. You can find the training script for WaveRNN `here `__. Please refer to :func:`torchaudio.pipelines.Tacotron2TTSBundle` for the usage. Example - "Hello world! T T S stands for Text to Speech!" .. image:: https://download.pytorch.org/torchaudio/doc-assets/TACOTRON2_WAVERNN_PHONE_LJSPEECH.png :alt: Spectrogram generated by Tacotron2 .. raw:: html Example - "The examination and testimony of the experts enabled the Commission to conclude that five shots may have been fired," .. image:: https://download.pytorch.org/torchaudio/doc-assets/TACOTRON2_WAVERNN_PHONE_LJSPEECH_v2.png :alt: Spectrogram generated by Tacotron2 .. raw:: html )4rS dataclassesrtypingrrrrrr ror torchaudio._internalr torchaudio.functionalr torchaudio.modelsr rtorchaudio.transformsrrrRr interfacer__all__rrrrBnnModuleVocoderr\rrrrrrrrrr_get_taco_params"TACOTRON2_GRIFFINLIM_CHAR_LJSPEECH__doc__#TACOTRON2_GRIFFINLIM_PHONE_LJSPEECH_get_wrnn_paramsTACOTRON2_WAVERNN_CHAR_LJSPEECH TACOTRON2_WAVERNN_PHONE_LJSPEECHrr+r#rs6 !:: 910=)  < ).<<)")/==):!ehhoo'9'A'A!0"*<*D*D"L'' ;;        $$$  -*N`     =/;Pb     %5 Tf     &6Vh   &DK,,r:&"!."*F'FI,,r:'#&/#+P#>S,,r:9**, # #+'J$@Q,,r:9**, $ ), (r+