i'ddlZddlZddlmZmZddlmZddlm Z m Z m Z ddl m Z ddlmZmZerddlZej$eZGdd e d ZGd d e Zd gZy)N) AudioInputmake_list_of_audio) BatchFeature)ProcessingKwargsProcessorMixinUnpack) TextInput)is_torch_availableloggingc*eZdZddidddddddd Zy ) MusicFlamingoProcessorKwargspaddingTi> max_length) sampling_ratereturn_attention_maskrptleft)return_tensors padding_side) text_kwargs audio_kwargs common_kwargsN)__name__ __module__ __qualname__ _defaults/var/www/vps2.regionflexible.com/Desarrollo/venv/lib/python3.12/site-packages/transformers/models/musicflamingo/processing_musicflamingo.pyrr(s2 t #%)# #"  IrrF)totalc eZdZdZ dfd ZdZdZdZ ddee ezde dzd e dzd e e d ef d Zed e efd ZxZS)MusicFlamingoProcessorah Constructs an MusicFlamingo processor which wraps an MusicFlamingo feature extractor and an MusicFlamingo tokenizer into a single processor. [`MusicFlamingoProcessor`] offers all the functionalities of [`WhisperFeatureExtractor`] and [`Qwen2TokenizerFast`]. See the [`~MusicFlamingoProcessor.__call__`] for more information. Args: feature_extractor ([`WhisperFeatureExtractor`]): The feature extractor is a required input. tokenizer ([`Qwen2TokenizerFast`]): The tokenizer is a required input. chat_template (`Optional[str]`, *optional*): The Jinja template to use for formatting the conversation. If not provided, the tokenizer's default chat template will be used. audio_token (`Optional[str]`, *optional*, defaults to `""`): Special token used to represent audio inputs in the chat template. audio_bos_token (`Optional[str]`, *optional*, defaults to `"<|sound_bos|>"`): Special token used to represent the beginning of audio. audio_eos_token (`Optional[str]`, *optional*, defaults to `"<|sound_eos|>"`): Special token used to represent the end of audio. max_audio_len (`int`, *optional*, defaults to 1200): Maximum length of audio sequences in seconds. Audio longer than this will be truncated. Nc||_|j||_||_t||||||_||_|j||_|j||_ y)N) chat_template) audio_tokenconvert_tokens_to_idsaudio_token_id max_audio_lensuper__init__audio_bos_tokenaudio_eos_tokenaudio_bos_token_idaudio_eos_token_id) selffeature_extractor tokenizerr%r&r,r-r) __class__s r r+zMusicFlamingoProcessor.__init__Sss''==kJ* *I]S.."+"A"A/"R"+"A"A/"Rrc2|dz dzdz}|dz dzdz}|S)Nr)r0 audio_lengthsconv_output_lengthsaudio_tokens_lengthss r _get_audio_token_lengthz.MusicFlamingoProcessor._get_audio_token_lengthfs2,q0Q6: 3a 7A=A##rctjtj|jd|Dcgc]}|jc}}|j |}t j t j|j}t|D]D\}} |j|j|j| zz|jz||||<F|Scc}w)N) torchstacksplitsumr:recompileescaper& enumeratesubr,r-) r0text padding_maskper_sample_windowssr7r9audio_token_patterni audio_lengths r _expand_audio_tokensz+MusicFlamingoProcessor._expand_audio_tokensks ekk,BRBRSUBVXj6k$lQUUW$lm #;;MJ jj43C3C)DE()=> OA|)--$$t'7'7,'FFI]I]]QDG   %msC,c`||jk(||jk(z||jk(zSN)r(r.r/)r0 input_idss r _get_audio_tokens_maskz-MusicFlamingoProcessor._get_audio_tokens_maskvs; $-- -D333 5D333 5 rrFaudio output_labelskwargsreturnc |jtfd|jji|}|d}|d}|j d}|dk7r"t |j jdt|tr|g}n3t|ttfrtd|Ds t di} |t|}t|t|k7r$t d t|d t|d t|d |j j"z} t|j$|j j"z} g} g} |D]}t|j&d }t)d|| zdz | z}|| kDr` placeholders in the text based on the post-pool frame counts of the audio windows, then tokenizes the provided strings as-is, and extracts log-mel features with [`WhisperFeatureExtractor`]. If `audio` is `None`, no audio processing is performed and the text is tokenized as-is (LM-only behavior). Args: text (`str` or `list[str]`): Input sequence or batch of sequences. audio (`np.ndarray` or `list[np.ndarray]`): Input audio or batch of audios as NumPy arrays. If provided, there must be as many `text` inputs as `audio` inputs. output_labels (bool, *optional*, default=False): Whether to return labels for training. Returns: [`BatchFeature`]: A dictionary with tokenized text (`input_ids`, `attention_mask`) and audio features (`input_features`, `input_features_mask`). tokenizer_init_kwargsrrrrz% only supports `return_tensors='pt'`.c3<K|]}t|tywrO) isinstancestr).0ts r z2MusicFlamingoProcessor.__call__..s9[QR*Q:L9[szAInvalid input text. Please provide a string, or a list of stringszGot z text but z audios; they must match 1:1.rrr5zAudio duration (z.1fz s) exceeds zs; truncating to first zs.attention_maskinput_features_maskrPilabels)data tensor_type) _merge_kwargsrr2 init_kwargsget ValueErrorr3rrYrZlisttupleallrlenintr1 chunk_lengthr)shapemaxloggerwarningappendminrangepoprMclonerQ pad_token_idr)r0rFrRrSrT call_kwargsrrr audio_inputs window_size max_windowsrH flat_chunksaudio_el n_samplesn_wintime_caprKstartendrG text_inputsrar`s r __call__zMusicFlamingoProcessor.__call__}s,:)d(( ( "&.."<"<  "-0 ">2 $)9: T ! 7 788]^_ _ dC 6DTD%=1c9[VZ9[6[`a a  &u-E4yCJ& 4D {*SZLHe!fggl?;d>T>T>a>aabKd00D4J4J4W4WWXK,. ,.K! <q 12 A K 7! ; KL;&NN*9|O7T+TUX*YYdeiewewdxyPQUQcQcPddfg(E"))%0y%+*=>uC&&xc':;< <"2411+NNL'++,<=L2>L. /,,TF4..v6 7<@F6T^^888 9#DN>BBrc|jj}|jj}ttj ||zdgzS)Nr_)r2model_input_namesr1rgdictfromkeys)r0 tok_names fea_namess r rz(MusicFlamingoProcessor.model_input_namessDNN44 **<< DMM)i"7;P:Q"QRSSr)Nzz <|sound_bos|>z <|sound_eos|>i)NF)rrr__doc__r+r:rMrQr rgrboolr rrrpropertyrZr __classcell__)r3s@r r#r#9s:''S&$  $(%* ]C$y/)]CD ]Cd{ ]C 56 ]C  ]C~T49TTrr#)rAnumpynp audio_utilsrrfeature_extraction_utilsrprocessing_utilsrrr tokenization_utils_baser utilsr r r= get_loggerrrorr#__all__rrr rsh, 94HH00   H %#35"gT^gTT $ $r