qi7dZddlZddlZddlZddlmZddlmZddl m Z ddl m Z m Z ddlmZd d lmZe j$eZe Gd d eZd gZy) z Processor class for Bark N) BatchFeature)ProcessorMixin) BatchEncoding)auto_docstringlogging) cached_file) AutoTokenizerceZdZddddZdfd Ze ddZ ddeffd Zdd e dzfd Z dd e dzfd Z e d efd ZddefdZe dd efdZxZS) BarkProcessorr semantic_prompt coarse_prompt fine_promptNc2t||||_y)a* speaker_embeddings (`dict[dict[str]]`, *optional*): Optional nested speaker embeddings dictionary. The first level contains voice preset names (e.g `"en_speaker_4"`). The second level contains `"semantic_prompt"`, `"coarse_prompt"` and `"fine_prompt"` embeddings. The values correspond to the path of the corresponding `np.ndarray`. See [here](https://suno-ai.notion.site/8b8e8749ed514b0cbf3f699013548683?v=bc67cff786b04b50b3ceb756fd05f68c) for a list of `voice_preset_names`. N)super__init__speaker_embeddings)self tokenizerr __class__s Z/opt/pipecat/venv/lib/python3.12/site-packages/transformers/models/bark/processing_bark.pyrzBarkProcessor.__init__*s #"4c T|jd}|t|||jdd|jdd|jdd|jdd|jdd||jd dddd }|:tj d t j j||d d}n,t|5}tj|}dddnd} d |vr||d <tj|fi|}|||S#1swY4xYw)a Instantiate a Bark processor associated with a pretrained model. Args: pretrained_model_name_or_path (`str` or `os.PathLike`): This can be either: - a string, the *model id* of a pretrained [`BarkProcessor`] hosted inside a model repo on huggingface.co. - a path to a *directory* containing a processor saved using the [`~BarkProcessor.save_pretrained`] method, e.g., `./my_model_directory/`. speaker_embeddings_dict_path (`str`, *optional*, defaults to `"speaker_embeddings_path.json"`): The name of the `.json` file containing the speaker_embeddings dictionary located in `pretrained_model_name_or_path`. If `None`, no speaker_embeddings is loaded. **kwargs Additional keyword arguments passed along to both [`~tokenization_utils_base.PreTrainedTokenizer.from_pretrained`]. tokenN subfolder cache_dirforce_downloadFproxieslocal_files_onlyrevision rrr r!r"rr# _raise_exceptions_for_gated_repo%_raise_exceptions_for_missing_entries'_raise_exceptions_for_connection_errors`z` does not exists , no preloaded speaker embeddings will be used - Make sure to provide a correct path to the json dictionary if wanted, otherwise set `speaker_embeddings_dict_path=None`. repo_or_path)rr) getr poploggerwarningospathjoinopenjsonloadr from_pretrained) cls!pretrained_processor_name_or_pathspeaker_embeddings_dict_pathkwargsrspeaker_embeddings_pathrspeaker_embeddings_jsonrs rr4zBarkProcessor.from_pretrained7sO, 7# ' 3&11, **[$7 **[$7%zz*:EB 9d3!',>!FJ5166;8= ' #'."'',,'HJfghi]` &*"12L6M)-3J)K&LL"&   )!335V">2!112S^W]^ Y;MNNLLs DD' push_to_hubc |j.tjtjj ||ddi}||d<|j D]}|j |}i} |j|D]m} tjtjj |d||d| || dtjj ||d| d | | <o| ||<ttjj ||d 5} tj|| dddt |4||fi|y#1swYxYw) a| Saves the attributes of this processor (tokenizer...) in the specified directory so that it can be reloaded using the [`~BarkProcessor.from_pretrained`] method. Args: save_directory (`str` or `os.PathLike`): Directory where the tokenizer files and the speaker embeddings will be saved (directory will be created if it does not exist). speaker_embeddings_dict_path (`str`, *optional*, defaults to `"speaker_embeddings_path.json"`): The name of the `.json` file that will contains the speaker_embeddings nested path dictionary, if it exists, and that will be located in `pretrained_model_name_or_path/speaker_embeddings_directory`. speaker_embeddings_directory (`str`, *optional*, defaults to `"speaker_embeddings/"`): The name of the folder in which the speaker_embeddings arrays will be saved. push_to_hub (`bool`, *optional*, defaults to `False`): Whether or not to push your model to the Hugging Face model hub after saving it. You can specify the repository you want to push to with `repo_id` (will default to the name of `save_directory` in your namespace). kwargs: Additional key word arguments passed along to the [`~utils.PushToHubMixin.push_to_hub`] method. Nv2T)exist_okr)_F) allow_picklez.npyw)rr.makedirsr/r0available_voice_presets_load_voice_presetnpsaver1r2dumprsave_pretrained) rsave_directoryr7speaker_embeddings_directoryr;r8embeddings_dict prompt_key voice_presettmp_dictkeyfprs rrHzBarkProcessor.save_pretrainedqsa8  " " . KK ^5QSWXcg h O.jCGG +N;=Y^h]iijknjo[p%S)%* %'GGLL1MR\Q]]^_b^ccgOh$iHSMj/7 + 7 bggll>3OPRUV /Z\ /2. /  FvF / /s EErMc ~|j|}i}|jd}dD]}||vrtd|d|dt|jjdd|||j dd|j d d|j d d |j d d|j d d ||j ddd d d  }|Mtdt j j|jjdd||d|dtj|||<|S)Nrr#Voice preset unrecognized, missing z% as a key in self.speaker_embeddings[z].r)/rrr Fr!r"r#r$r(z{` does not exists , no preloaded voice preset will be used - Make sure to provide correct paths to the z embeddings.) rr* ValueErrorr r+r.r/r0rEr3)rrMr8voice_preset_pathsvoice_preset_dictrrOr/s rrDz BarkProcessor._load_voice_presetsi!44\B 7#F 3C,, 9#>cdpcqqst''++NC@"3' **[$7 **[$7%zz*:EB 9d3!',>!FJ5166;8= D| "'',,t'>'>'B'B>SV'WYkloYpqrsjjviwx # &(WWT] c "7 3:! rc ^dD]}||vrtd|dt||tjs't |dt |j |dt||j|j |k7st|dt |j |dy)NrrRz as a key.z voice preset must be a z D ndarray.) rT isinstancerEndarray TypeErrorstr preset_shapelenshape)rrMrOs r_validate_voice_preset_dictz)BarkProcessor._validate_voice_preset_dictsF jC,& #Fse:!VWWl3/<3%'?DDUDUVYDZ@[?\\f ghh<$**+t/@/@/EE C5(@TEVEVWZE[A\@]]g!hii jrreturnc|jgSt|jj}d|vr|jd|S)z Returns a list of available voice presets. Returns: `list[str]`: A list of voice preset names. r))rlistkeysremove)r voice_presetss rrCz%BarkProcessor.available_voice_presetssJ  " " *IT4499;< ] *   0rremove_unavailablecFg}|js|jD]%} |j|}|j |'|r%t jdt|d|d|r|D]}|j|=yyy#t$r|j |YwxYw)NzThe following z' speaker embeddings are not available: zU If you would like to use them, please check the paths or try downloading them again.) rrCrDrTappendr_r,r-r])rrfunavailable_keysrMrVs r_verify_speaker_embeddingsz(BarkProcessor._verify_speaker_embeddingss  " " . $ < < D (,(?(? (M% 001BC D $S)9%:$;;bcsbtukk "$4>L// =>"! /"$++L9sBB B c |t|tswt|tr,|j ||jvr|j |}n;t|tr|j ds|dz}t j|}| |j|fi|t||}|j|f|d||||d|} ||| d<| S)a voice_preset (`str`, `dict[np.ndarray]`): The voice preset, i.e the speaker embeddings. It can either be a valid voice_preset name, e.g `"en_speaker_1"`, or directly a dictionary of `np.ndarray` embeddings for each submodel of `Bark`. Or it can be a valid file name of a local `.npz` single voice preset containing the keys `"semantic_prompt"`, `"coarse_prompt"` and `"fine_prompt"`. Returns: [`BatchEncoding`]: A [`BatchEncoding`] object containing the output of the `tokenizer`. If a voice preset is provided, the returned object will include a `"history_prompt"` key containing a [`BatchFeature`], i.e the voice preset with the right tensors type. z.npz)data tensor_type max_length)return_tensorspaddingrnreturn_attention_maskreturn_token_type_idsadd_special_tokenshistory_prompt) rXdictr[rrDendswithrEr3r_rr) rtextrMrornrsrqrrr8 encoded_texts r__call__zBarkProcessor.__call__s0  #J|T,J<-++7 D$;$;;#66|D lC09N9Nv9V#/L!ww|4  # ,D , ,\ DV D'\~VL%t~~   ) !"7"71      #-9L) *r)N)speaker_embeddings_path.json)rzrF)T)NNptFTF)__name__ __module__ __qualname__r\r classmethodr4boolrHr[rDrur_propertyrbrCrjrrry __classcell__)rs@rr r "sL 5Mk7O7Ox&D%9! 6G  6Gp"!sTz"!H jt j   >T>. "#7 77rr )__doc__r2r.numpyrEfeature_extraction_utilsrprocessing_utilsrtokenization_utils_baserutilsrr utils.hubr autor get_loggerr}r,r __all__rrrsd 4.4,$    H %TNTTn  r