o iB+@sddlZddlmZddlZddlZddlmZmZm Z m Z m Z ddl m Z mZer1ddlmZe eZded ed ejfd d Zeed dGddde ZdS)N)Any)add_end_docstringsis_torch_availableis_torchaudio_availableis_torchcodec_availablelogging)Pipelinebuild_pipeline_init_args),MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING_NAMESbpayload sampling_ratereturnc Cs|}d}d}dddd|d|d|d d d d g }z tj|tjtjd }Wn ty-tdw||}|d}t|tj} | j ddkrItd| S)z? Helper function to read an audio file through ffmpeg. 1f32leffmpegz-izpipe:0z-acz-arz-fz -hide_bannerz -loglevelquietzpipe:1)stdinstdoutzFffmpeg was not found but is required to load audio files from filenamerzMalformed soundfile) subprocessPopenPIPEFileNotFoundError ValueError communicatenp frombufferfloat32shape) r raracformat_for_conversionffmpeg_commandffmpeg_process output_stream out_bytesaudior(m/sda-disk/www/egybert/egybert_env/lib/python3.10/site-packages/transformers/pipelines/audio_classification.py ffmpeg_reads8  r*T)has_feature_extractorc seZdZdZdZdZdZdZfddZde j e Be Be Bdedee e efffd d Zdd d ZddZddZdddZZS)AudioClassificationPipelinea Audio classification pipeline using any `AutoModelForAudioClassification`. This pipeline predicts the class of a raw waveform or an audio file. In case of an audio file, ffmpeg should be installed to support multiple audio formats. Example: ```python >>> from transformers import pipeline >>> classifier = pipeline(model="superb/wav2vec2-base-superb-ks") >>> classifier("https://huggingface.co/datasets/Narsil/asr_dummy/resolve/main/1.flac") [{'score': 0.997, 'label': '_unknown_'}, {'score': 0.002, 'label': 'left'}, {'score': 0.0, 'label': 'yes'}, {'score': 0.0, 'label': 'down'}, {'score': 0.0, 'label': 'stop'}] ``` Learn more about the basics of using a pipeline in the [pipeline tutorial](../pipeline_tutorial) This pipeline can currently be loaded from [`pipeline`] using the following task identifier: `"audio-classification"`. See the list of available models on [huggingface.co/models](https://huggingface.co/models?filter=audio-classification). FTcsNd|vr|ddurd|d<nd|vrd|d<tj|i||tdS)Ntop_k)super__init__check_model_typer )selfargskwargs __class__r(r)r0bs  z$AudioClassificationPipeline.__init__inputsr4rc stj|fi|S)a Classify the sequence(s) given as inputs. See the [`AutomaticSpeechRecognitionPipeline`] documentation for more information. Args: inputs (`np.ndarray` or `bytes` or `str` or `dict`): The inputs is either : - `str` that is the filename of the audio file, the file will be read at the correct sampling rate to get the waveform using *ffmpeg*. This requires *ffmpeg* to be installed on the system. - `bytes` it is supposed to be the content of an audio file and is interpreted by *ffmpeg* in the same way. - (`np.ndarray` of shape (n, ) of type `np.float32` or `np.float64`) Raw audio at the correct sampling rate (no further check will be done) - `dict` form can be used to pass raw audio sampled at arbitrary `sampling_rate` and let this pipeline do the resampling. The dict must be either be in the format `{"sampling_rate": int, "raw": np.array}`, or `{"sampling_rate": int, "array": np.array}`, where the key `"raw"` or `"array"` is used to denote the raw audio waveform. top_k (`int`, *optional*, defaults to None): The number of top labels that will be returned by the pipeline. If the provided number is `None` or higher than the number of labels available in the model configuration, it will default to the number of labels. function_to_apply(`str`, *optional*, defaults to "softmax"): The function to apply to the model output. By default, the pipeline will apply the softmax function to the output of the model. Valid options: ["softmax", "sigmoid", "none"]. Note that passing Python's built-in `None` will default to "softmax", so you need to pass the string "none" to disable any post-processing. Return: A list of `dict` with the following keys: - **label** (`str`) -- The label predicted. - **score** (`float`) -- The corresponding probability. )r/__call__)r2r7r4r5r(r)r8ls"z$AudioClassificationPipeline.__call__NcKsxi}|dur|jjj|d<n||jjjkr|jjj}||d<|dur3|dvr.td|d||d<nd|d<ii|fS)Nr-)softmaxsigmoidnonez'Invalid value for `function_to_apply`: z2. Valid options are ['softmax', 'sigmoid', 'none']function_to_applyr9)modelconfig num_labelsr)r2r-r<r4postprocess_paramsr(r(r)_sanitize_parameterss    z0AudioClassificationPipeline._sanitize_parametersc Cst|tr1|ds|drtj|ddj}nt|d }|}Wdn1s,wYt|tr=t ||j j }t rPddl }t||jrP|}troddl }ddl}t||jjro|}|j}||jd}t|tr|}d|vrd |vsd |vstd |d d}|dur|d d|d d}|d}|}||j j krddl }trdd lm} ntd| t|t!j"r|#|n|||j j }t|t!j"st$dt%|j&dkrtd|j ||j j dd} |j'dur| j(|j'd} | S)Nzhttp://zhttps://T)follow_redirectsrbr)arrayrrrawrDzWhen passing a dictionary to AudioClassificationPipeline, the dict needs to contain a "raw" key containing the numpy array or torch tensor representing the audio and a "sampling_rate" key, containing the sampling_rate associated with that arraypath) functionalztorchaudio is required to resample audio samples in AudioClassificationPipeline. The torchaudio package can be installed through: `pip install torchaudio`.z2We expect a numpy ndarray or torch tensor as inputr zFWe expect a single channel audio input for AudioClassificationPipelinept)rreturn_tensors)dtype)) isinstancestr startswithhttpxgetcontentopenreadbytesr*feature_extractorrrtorchTensorcpunumpyr torchcodecdecoders AudioDecoderget_all_samplesdata sample_ratedictcopyrpopr torchaudiorG ImportErrorresamplerndarray from_numpy TypeErrorlenrrJto) r2r7frUrY_audio_samples_array_inputsin_sampling_rateF processedr(r(r) preprocesssn                z&AudioClassificationPipeline.preprocesscCs|jdi|}|S)Nr()r=)r2 model_inputs model_outputsr(r(r)_forwardsz$AudioClassificationPipeline._forwardr.r9csv|dkr |jdd}n|dkr|jd}n|jd}||\}}|}|}fddt||D}|S)Nr9rr:cs$g|]\}}|jjj|dqS))scorelabel)r=r>id2label).0rv_idr2r(r) s$z;AudioClassificationPipeline.postprocess..)logitsr9r:topktolistzip)r2rsr-r<probsscoresidslabelsr(r{r) postprocesss z'AudioClassificationPipeline.postprocess)NN)r.r9)__name__ __module__ __qualname____doc___load_processor_load_image_processor_load_feature_extractor_load_tokenizerr0rrerSrLr_rlistr8rArqrtr __classcell__r(r(r5r)r,Bs 4 $Kr,)rtypingrrNrXrutilsrrrrrbaser r models.auto.modeling_autor get_loggerrloggerrSintrer*r,r(r(r(r)s    $