+ jT.^RIt^RIt^RIt^RIHuHt^RIH t ^RI H t ^RI H t ^RIHtHtHt^RIHt]P(!]4tR RRllt]!R R 7!R R ] 44tR .tR#)N)mel_filter_bank)SequenceFeatureExtractor) BatchFeature)PaddingStrategy TensorTypelogging)requirescHV^8dQhR\R\R\R\/#)valuename tolerancereturn)floatstrint)formats"w/home/luca/screen-checking/.venv/lib/python3.14/site-packages/transformers/models/inkling/feature_extraction_inkling.py __annotate__rs(cescz\V4p\W, 4V8d\V RV 24h\V4#)z. must resolve to an integer sample count, got )roundabs ValueErrorr)r r rroundeds&&& r _to_exact_intrs<ElG 5?i'D6!OPUwWXX w<r)backendsczaa]tRt^%toRtRR.tR V3RlV3RllltR V3RlRlltR V3RlR lltR t Vt V;t #)InklingFeatureExtractora Constructs a TML audio feature extractor, which converts raw audio waveforms into log-mel spectrogram features (mel filterbank energies in log10 space). The quantization of these features into discrete dMel bins is performed downstream by [`InklingProcessor`]. This feature extractor inherits from [`~feature_extraction_sequence_utils.SequenceFeatureExtractor`] which contains most of the main methods. Users should refer to this superclass for more information regarding those methods. Args: feature_size (`int`, *optional*, defaults to 80): The feature dimension of the extracted features, i.e. the number of mel filterbanks. sampling_rate (`int`, *optional*, defaults to 16000): The sampling rate at which the audio files should be digitized, expressed in hertz (Hz). padding_value (`float`, *optional*, defaults to 0.0): The value used to pad the log-mel spectrograms to the same length in a batch. audio_token_duration_s (`float`, *optional*, defaults to 0.05): Duration, in seconds, represented by a single audio token, i.e. the STFT hop length. window_size_multiplier (`float`, *optional*, defaults to 2.0): Multiplier applied to `audio_token_duration_s` to obtain the STFT window length. n_fft (`int`, *optional*): FFT size. Defaults to the window length (`audio_token_duration_s * window_size_multiplier * sampling_rate`) when not provided. input_featuresinput_features_maskc L<V^8dQhRS[RS[RS[RS[RS[RS[R,/#)r feature_size sampling_rate padding_valueaudio_token_duration_swindow_size_multipliern_fftN)rr)r __classdict__s"rr$InklingFeatureExtractor.__annotate__BsS+c+c+c+c +c !& +c !& +cTz+crc <\S V`!RRVRVRV/VBW@nWPn\ WB,R4Vn\ WE,V,R4VnT;'g VP VnVP ^8:g#VP ^8:gVP^8:d \R4h\P!VP R\PR7Vn \VP^,^,VRVR , VR R R 7p\P!\P !VP"\PR 74VnR #)r#r$r%z&audio_token_duration_s * sampling_ratez?audio_token_duration_s * window_size_multiplier * sampling_ratez7hop_length, window_size, and n_fft must all be positiveT)periodicdtype@slaney)num_frequency_binsnum_mel_filters min_frequency max_frequencyr$norm mel_scale)r-N)super__init__r&r'r hop_length window_sizer(rtorch hann_windowfloat32windowr from_numpynpascontiguousarrayT mel_filters) selfr#r$r%r&r'r(kwargsrD __class__s &&&&&&&, rr9 InklingFeatureExtractor.__init__Bs>  % ' (   '=#&<#' " 24\ ) " ;m K M ..d.. ??a 4#3#3q#8DJJ!OVW W''(8(84u}}] &#zzQ2('#-' !++B,@,@VXV`V`,abrcT<V^8dQhRS[PRS[RS[P/#)r waveformdevicer)r<Tensorr)rr)s"rrr*os-((ell(C(\a\h\h(rc ZVPV4p\P!VPR,VP, 4VP,VPR,, p\ VP VP, ^4p\P!WV34p\P!VVP VPVPVPPV4RRR7p\P!V4pVP^4PR4P!R4P#4pVP$PV4pWv,pVP!R4P'4pVP)^^4#)FT)r: win_lengthr?centerreturn_complexg|=)tomathceilshaper:maxr(Fpadr<stftr;r? view_as_realpowsum clamp_minsqrtrDlog10 transpose) rErJrK right_padleft_padrZ magnitudesrDmel_specs &&& r_torch_extract_fbank_features5InklingFeatureExtractor._torch_extract_fbank_featuresos@;;v&IIhnnR04??BCdooUX`XfXfgiXjj tzzDOO3Q755i#89zz  JJ'';;>>&) ''- ^^A&**2.88?DDF &&))&1 +%%e,224!!!Q''rcr<V^8dQhRS[PS[S[,,S[S[P,,S[S[S[,,,RS[R,RS[S[,S[,RS[R,RS[RS[R,RS[R,R S[S[,R,R S[R,R S[ / #) r raw_speechr$Npadding max_length truncationpad_to_multiple_ofreturn_attention_maskreturn_tensorsrKr) rAndarraylistrrboolrrrr)rr)s"rrr*snCnCJJe,tBJJ/??$tE{BSSnCTznCo- nC $J nC  nC $JnC $d{nCj(4/nCd nC nCrc aVe;W P8wd*\RVP RVP RV R24hM\PR4VPP oRV3Rllp \ V\P4'd\P!V4p\ V\P4'd/VP^8d\RVP R 24hV.p M\ V\\34'dm\V4^8Xd \R 4h\ V^,\ \"\P$\P&34'dV.p M&\V4p M\)R S R \+V4 24hV U u.uFq!V 4R,NK pp VUu.uFp\V4NK pp\-R VRV/4pVP/VVVVVRR7pVP0P3R4pVP5VV 4p\P6!VP8P;V 4VP<,^, VP<RR7p\P>!VP@^,V R7R,VR,8pVVPCR4,pR V/pV'dVVR&\-VVR7#uup iuupi)ay Extract log-mel spectrogram features from one or several audio clip(s). Args: raw_speech (`np.ndarray`, `list[float]`, `list[np.ndarray]`, `list[list[float]]`): The sequence or batch of sequences to be padded. Each sequence can be a numpy array, a list of float values, a list of numpy arrays or a list of list of float values. Must be mono channel audio at `self.sampling_rate`, not stereo, i.e. single float per timestep. Decoding and resampling of raw audio (bytes / paths / URLs) is handled upstream by the processor's `apply_chat_template`, not here. sampling_rate (`int`, *optional*): The sampling rate of `raw_speech`, used only to validate against `self.sampling_rate`. device (`str`, *optional*, defaults to `"cpu"`): The device on which the log-mel spectrogram is computed in `_torch_extract_fbank_features`. zWThe model corresponding to this feature extractor was trained using a sampling rate of zB. Please make sure that the provided audio input was sampled with z and not .zIt is strongly recommended to pass the `sampling_rate` argument to this function. Failing to do so can result in silent errors that might be hard to debug.c<V^8dQhRRR\P/#)r clipz np.ndarray | torch.Tensor | listr)r<rL)rs"rr6InklingFeatureExtractor.__call__..__annotate__s  = %,, rc<\V\P4'dTM)\P!\P !V44pVP \P4pVP^8Xd.\PRS R24VPRR7pV#VP^8wd#\R\VP4 R24hV#)r z2Only mono-channel audio is supported for input to zB. Taking the mean over the channel (last) axis to convert to mono.)dimzDEach audio clip must be 1-D (mono) or 2-D (multichannel), got shape rtrR) isinstancer<rL as_tensorrAasarrayrSr>ndimlogger warning_oncemeanrtuplerV)rvtensorcls_names& r_to_mono2InklingFeatureExtractor.__call__.._to_monos'ell;;TQSQ[Q[\`QaAbFYYu}}-F{{a##H SWW , M ! Z[`agamam[nZoopqMrzCA single array input must be 1-D (mono) or 2-D (multichannel); got z2 dims. Pass a list of arrays for a batch of clips.zReceived an empty audio input.z!Unsupported audio input type for z: r audio_lengthspt)rjrkrlrmrofloor) rounding_mode)rKr!)data tensor_type)NNNNrR)Nr)"r$rr~rrG__name__rzrArpr<r@rLr}rqrlenrrintegerfloating TypeErrortyperrYr squeezerfdivrrSr:arangerV unsqueeze)rErir$rjrkrlrmrnrorKrFrclipsrvspeechrbatched_speech padded_inputsinput_waveformsr num_framesr!rrs&&&&&&&&&&, @r__call__ InklingFeatureExtractor.__call__s8  $ 2 22 #1123((,(:(:';9]OSTV3   \  >>**   j"** - -))*5J j%,, / /" YZdZiZiYjkBB LE  T5M 2 2:!# !ABB*Q-#ubjj"++)NOO# Z(?zDQ[L\K]^_ _:?@%$htnW--% @4>>:V: >%'7_Vc&de !!1 ! (66>>rB;;OVTYY  ' ' * *6 2T__ Dq H$//ip $ll>+?+?+B6RSZ[^hip^qq'*=*G*G*KK .1 +>D& '>BB?A?s K(;K-)r&r:rDr(r?r;r')Pi>r.g?r/N)cpu)NTNFNTNr) r __module__ __qualname____firstlineno____doc__model_input_namesr9rfr__static_attributes____classdictcell__ __classcell__)rGr)s@@rrr%sD2*+@A+c+cZ((2nCnCnCrr)gư>)r<)rTnumpyrAr<torch.nn.functionalnn functionalrX audio_utilsr!feature_extraction_sequence_utilsrfeature_extraction_utilsrutilsrrrutils.import_utilsr get_loggerrr~rr__all__r7rrrsq  *I499*   H % :PC6PCPCf % %r