+ jBRt^RIt^RIHt^RIHtHtHt^RI H t ^RI H t ^RI HtHt]!4'd^RIt]P"!]4t!RR] 4tR.tR#) z% Feature extractor class for Whisper N)is_torch_available)mel_filter_bank spectrogramwindow_function)SequenceFeatureExtractor) BatchFeature) TensorTypeloggingcaa]tRt^!toRtR.tR V3RlltV3RlRltRV3RlRllt] RV3RlR ll4t RV3R lR llt R t Vt V;t#)WhisperFeatureExtractora/ Constructs a Whisper feature extractor. 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. This class extracts mel-filter bank features from raw speech using a custom numpy implementation of the `Short Time Fourier Transform` which should match pytorch's `torch.stft` equivalent. Args: feature_size (`int`, *optional*, defaults to 80): The feature dimension of the extracted features. sampling_rate (`int`, *optional*, defaults to 16000): The sampling rate at which the audio files should be digitalized expressed in hertz (Hz). hop_length (`int`, *optional*, defaults to 160): Length of the overlapping windows for the STFT used to obtain the Mel Frequency coefficients. chunk_length (`int`, *optional*, defaults to 30): The maximum number of chunks of `sampling_rate` samples used to trim and pad longer or shorter audio sequences. n_fft (`int`, *optional*, defaults to 400): Size of the Fourier transform. padding_value (`float`, *optional*, defaults to 0.0): Padding value used to pad the audio. Should correspond to silences. dither (`float`, *optional*, defaults to 0.0): Adds dithering. In other words, adds a small Gaussian noise to each frame. E.g. use 0.0001 to add dithering with a normal distribution centered around 0.0 with standard deviation 0.0001 (assuming [-1,+1] range of raw_speech). The value 0.0 means no dithering. Dithering has similar effect as `spectrogram(mel_floor=...)`. It reduces the high log_mel_fbank values for signals with hard-zero sections, when VAD cutoff is present in the signal. input_featuresc <\S V`!R RVRVRVRV/V BWPnW0nW@nWB,VnVP V,VnW nWpn\^V^,,VRRVRRR7Vn R#) feature_size sampling_rate padding_valuereturn_attention_maskg@@slaney)num_frequency_binsnum_mel_filters min_frequency max_frequencyrnorm mel_scaleN) super__init__n_fft hop_length chunk_length n_samples nb_max_framesrditherr mel_filters) selfrrrrrrr"rkwargs __class__s &&&&&&&&&,w/home/luca/screen-checking/.venv/lib/python3.14/site-packages/transformers/models/whisper/feature_extraction_whisper.pyr WhisperFeatureExtractor.__init__Es  % ' ( #8     $(%5!^^z9* * 5A:~( ' cT<V^8dQhRS[PRS[RS[P/#)waveform_batchdevicereturnnpndarraystr)format __classdict__s"r' __annotate__$WhisperFeatureExtractor.__annotate__is-SUWU_U_r)c  VR8wd\RV R24h.pVFp\V\VPR4VPVPRVP VP RR7pVRR R 13,p\P!WUP4R , 4pVR ,R , pVPV4K \P!V4pV#) z Compute the log-mel spectrogram of the provided audio, gives similar results to Whisper's original torch implementation with 1e-5 tolerance. cpuz Got device `z` for feature extraction, but feature extraction on CUDA accelerator devices requires torch, which is not installed. Either set `device='cpu'`, or install torch according to the official instructions: https://pytorch.org/get-started/locally/hanng@log10) frame_lengthrpowerr"r#log_melNNNN @@) ValueErrorrrrrr"r#r0maximummaxappendarray)r$r,r-log_spec_batchwaveformlog_specs&&& r'_np_extract_fbank_features2WhisperFeatureExtractor._np_extract_fbank_featuresis U?vh'qq  &H" F3!ZZ??{{ ,, H 3B3'Hzz(LLNS,@AH 3#-H  ! !( +'.1r)cT<V^8dQhRS[PRS[RS[P/#)r+rHr-r.r/)r3r4s"r'r5r6s-! ! bjj! #! Z\ZdZd! r)c \P!V4PV\P4p\P!VP VR7pVP R8wdOWP \P!VPVPVPR7,, p\P!WP VPVRR7pVRRR13,P4^,P4p\P!VP4PV\P4pVP V,p\P"!VRR 7P%4pVP'4^8XdQVP)^RR 7^,P)^RR 7^,p \P*!WR , 4pM+\P*!WP)4R , 4pVR ,R , pVR 8wdVP-4P/4pVP14#)z Compute the log-mel spectrogram of the audio using PyTorch's GPU-accelerated STFT implementation with batching, yielding results similar to cpu computing with 1e-5 tolerance. )r-r)dtyper-T)windowreturn_complex.Ng|=)min)dimkeepdimr?r@r8rA)torch from_numpytofloat32 hann_windowrr"randnshaperNr-stftrabs contiguousr#Tclampr:rRrDrCdetachr8numpy) r$rHr-rOr[ magnitudesr#mel_specrImax_vals &&& r'_torch_extract_fbank_features5WhisperFeatureExtractor._torch_extract_fbank_featuress ##H-00G""4::f= ;;#   ekk(.._g_n_n&oo oHzz(JJ_cd 38n((*a/;;= &&t'7'78;;FEMMR ==:-;;xU399; <<>Q llq$l7:>>1d>STUVG}}X}=H}}X||~/CDHsNc) U?(,,.H~~r)c <V^8dQhRS[S[P,RS[S[P,RS[RS[S[P,/#)r+ input_valuesattention_maskrr.)listr0r1float)r3r4s"r'r5r6sH##2::&#8rA) r0rFint32zipsummeansqrtvarrZrE)rhrirnormed_input_valuesvectorlength normed_slicexs&&& r'zero_mean_unit_var_norm/WhisperFeatureExtractor.zero_mean_unit_var_norms  %XXnbhh?N"$ "%l4F4Fr4J"K &)=)=)? ?2776RYSY?K^K^K`cgKgChh ..q11,9)#**<8 #L#"Vb"bUaPQLBGGAEEGdN4K#K#KUa "b""#cs AD ct<V^8dQhRS[PS[S[,,S[S[P,,S[S[S[,,,RS[RS[R,RS[S[,R,RS[R,RS[R,RS[R,R S[R,R S[R,R S[R,R S[/ #) r+ raw_speech truncationpad_to_multiple_ofNreturn_tensorsrpadding max_lengthr do_normalizer-r.) r0r1rjrkboolintr2rr)r3r4s"r'r5r6sYYJJe,tBJJ/??$tE{BSSYY $J Y j(4/ Y $d{ YtY$JYTzYTkYd Y Yr)c VeRWP8wdA\RVPP RVP RVP RV R2 4hM-\P RVPP R24\ V\P4;'d\VP4^8p V 'd)\VP4^8d\R V 24hT ;'gK\ V\\34;'d-\ V^,\P\\34p V 'dBVUu.uF4p\P!V.\PR 7PNK6 ppMV 'gH\ V\P4'g(\P!V\PR 7pMq\ V\P4'dRVP \P !\P"4Jd VP%\P4pV 'g#\P!V.4P.p\'R V/4pTP)TTV'dTM VP*TTT;'gT R 7pV 'dRVP-VR ,VR ,VP.R7VR &\P0!VR ,^R7VR &VP3R 4P5^^^4p\74'd VP8M VP:pV!V^,V 4p\ V^,\4'd;VUu.uF)p\P!V\PR 7NK+ upVR &MVVR &V'dbVR ,RRRVP<13,pVR ,P^,VP<,^8wdVRRR13,pVVR &VeVP?V4pV#uupiuupi)a Main method to featurize and prepare for the model one or several sequence(s). Implementation uses PyTorch for the STFT computation if available, otherwise a slower NumPy based one. 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, not stereo, i.e. single float per timestep. truncation (`bool`, *optional*, default to `True`): Activates truncation to cut input sequences longer than *max_length* to *max_length*. pad_to_multiple_of (`int`, *optional*, defaults to None): If set will pad the sequence to a multiple of the provided value. This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128. return_tensors (`str` or [`~utils.TensorType`], *optional*): If set, will return tensors instead of list of python integers. Acceptable values are: - `'pt'`: Return PyTorch `torch.Tensor` objects. - `'np'`: Return Numpy `np.ndarray` objects. return_attention_mask (`bool`, *optional*): Whether to return the attention mask. If left to the default, will return the attention mask according to the specific feature_extractor's default. [What are attention masks?](../glossary#attention-mask) For Whisper models, `attention_mask` should always be passed for batched inference, to avoid subtle bugs. padding (`str` or [`~utils.PaddingStrategy`], *optional*, defaults to `'max_length'`): Activates and controls padding. Accepts the following values: - `'longest'`: Pad to the longest sequence in the batch (or no padding if only a single sequence is provided). - `'max_length'` (default): Pad to a maximum length specified with the argument `max_length` or to the maximum acceptable input length for the model if that argument is not provided. - `'do_not_pad'`: No padding (i.e., can output a batch with sequences of different lengths). max_length (`int`, *optional*): Controls the maximum length to use by one of the truncation/padding parameters. If left unset or set to `None`, this will use the predefined model maximum length if a maximum length is required by one of the truncation/padding parameters. If the model has no specific maximum input length (like XLNet) truncation/padding to a maximum length will be deactivated. sampling_rate (`int`, *optional*): The sampling rate at which the `raw_speech` input was sampled. It is strongly recommended to pass `sampling_rate` at the forward call to prevent silent errors and allow automatic speech recognition pipeline. do_normalize (`bool`, *optional*, defaults to `False`): Whether or not to zero-mean unit-variance normalize the input. Normalizing can help to significantly improve the performance of the model. device (`str`, *optional*, defaults to `'cpu'`): Specifies the device for computation of the log-mel spectrogram of audio signals in the `_torch_extract_fbank_features` method. (e.g., "cpu", "cuda") **kwargs: Not supported by WhisperFeatureExtractor.__call__() and ignored. Nz3The model corresponding to this feature extractor: z& was trained using a sampling rate of zI. Please make sure that the provided `raw_speech` input was sampled with z and not .zDIt is strongly recommended to pass the `sampling_rate` argument to `zN()`. Failing to do so can result in silent errors that might be hard to debug.z2Only mono-channel audio is supported for input to )rNr )rrr|r}rri)rir)axisr>rA) rrBr&__name__loggerwarning isinstancer0r1lenrZrjtupleasarrayrWr^rNfloat64astyperpadr rxrstackget transposerrerJrconvert_to_tensors)r$r{r|r}r~rrrrrr-r%is_batched_numpy is_batchedspeechbatched_speech padded_inputsr extract_fbank_featuresfeaturerescaled_attention_masks&&&&&&&&&&&, r'__call__ WhisperFeatureExtractor.__call__sP  $ 2 22 I$..JaJaIbc))-););(<=))-););("00/K/M* + /1hh}EU7V]^._M* +'**+;<FFq!QO3E2F2FD . .DLkLk 0q0A6J nQ' . .dr.sdrY`rzz'/Tdr.sM* +/=M* + &34D&EaI[DOOI[F[&\ # -.44Q7$//IQN*A!SbS&*I'.EM* +  %)<<^LMy]R/ts :Q /Q )rr"rr#rr r!r)Pi>irrF)r8)r) TNNNrNNNr8)r __module__ __qualname____firstlineno____doc__model_input_namesrrJre staticmethodrxr__static_attributes____classdictcell__ __classcell__)r&r4s@@r'r r !s\B**" HrsY"HHI4(   H %y6yx % %r)