o ۷i@sxdZddlZddlmZmZddlZddlmZm Z m Z m Z ddl m Z ddlmZeeZGdd d e Zd gZdS) zXcodec model configurationN)OptionalUnion) AutoConfig DacConfig HubertConfig WavLMConfig)PretrainedConfig)loggingcseZdZdZdZeedZdddddgddgddgdddd ddf d ee e d e d e d e e de e de e de de dee de de e efde e efffdd Zede fddZede fddZede fddZede fdd Zede fd!d"Zede fd#d$ZZS)% XcodecConfiga This is the configuration class to store the configuration of an [`XcodecModel`]. It is used to instantiate a Xcodec model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the [Manel/X-Codec](https://huggingface.co/Manel/X-Codec) architecture. Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PretrainedConfig`] for more information. Args: target_bandwidths (`List[float]`, *optional*, defaults to `[0.5, 1, 1.5, 2, 4]`): The range of different bandwidths (in kbps) the model can encode audio with. sample_rate (`int`, *optional*, defaults to 16000): The sampling rate at which the audio waveform should be digitalized, in hertz (Hz). kernel_size (`int`, *optional*, defaults to 3): Kernel size for the initial semantic convolution. channel_ratios (`List[float]`, *optional*, defaults to `[1, 1]`): Expansion factors for the number of output channels in each semantic block. strides (`List[int]`, *optional*, defaults to `[1, 1]`): Strides for each semantic encoder block. block_dilations (`List[int]`, *optional*, defaults to `[1, 1]`): Dilation factors for the residual units in semantic blocks. unit_kernel_size (`int`, *optional*, defaults to 3): Kernel size inside each ResidualUnit in semantic blocks. codebook_size (`int`, *optional*, defaults to 1024): Number of entries in each residual quantizer's codebook. codebook_dim (`int`, *optional*): Dimensionality of each codebook vector. Defaults to sum of hidden size of acoustic and semantic models. initializer_range (`float`, *optional*, defaults to 0.02): Standard deviation of the truncated normal initializer for all weight matrices. acoustic_model_config (`Union[Dict, DacConfig]`, *optional*): An instance of the configuration for the acoustic (DAC) model. semantic_model_config (`Union[Dict, HubertConfig, WavLMConfig]`, *optional*): An instance of the configuration object for the semantic (HuBERT) model. Example: ```python >>> from transformers import XcodecModel, XcodecConfig >>> # Initializing configuration >>> configuration = XcodecConfig() >>> # Initializing a model (with random weights) from the configuration >>> model = XcodecModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```xcodec)acoustic_model_configsemantic_model_configNi>rg{Gz?target_bandwidths sample_rate kernel_sizechannel_ratiosstridesblock_dilationsunit_kernel_size codebook_size codebook_diminitializer_ranger rc  s`tjd i| | durtdgddgddd|_n t| tr*td i| |_nt| tr3| |_n tdt| | durEt|_ n7t| tred| vrWt | d|_ n%t dtd i| |_ nt| tsot| trs| |_ n td t| |durgd }||_||_||_||_||_||_||_||_| |_| dur|jj|j j} | |_dS) N@)r)encoder_hidden_sizedownsampling_ratiosdecoder_hidden_sizeupsampling_ratios hidden_sizezDacoustic_model_config must be a dict or DacConfig instance, but got _name_or_pathz_Could not determine semantic model type from config architecture. Defaulting to `HubertConfig`.zUsemantic_model_config must be a dict, HubertConfig, or WavLMConfig instance, but got )g?rg?rr)super__init__rr isinstancedict ValueErrortyperrrfrom_pretrainedloggerwarningrrrrrrrrrrr%r)selfrrrrrrrrrrr rkwargs __class__r'e/home/ubuntu/vllm_env/lib/python3.10/site-packages/transformers/models/xcodec/configuration_xcodec.pyr)YsX       zXcodecConfig.__init__returncCst|j|jSN)mathceilr hop_lengthr1r'r'r5 frame_ratezXcodecConfig.frame_ratecCs|jjSr7)rr%r;r'r'r5semantic_hidden_sizesz!XcodecConfig.semantic_hidden_sizecCstt|jjSr7)intnpprodr r"r;r'r'r5r:r=zXcodecConfig.hop_lengthcCstt|jSr7)r8r9log2rr;r'r'r5codebook_nbitsr=zXcodecConfig.codebook_nbitscCs|jj|jjSr7)r r%rr;r'r'r5r%szXcodecConfig.hidden_sizecCstd|jd|j|jS)Ni)r?rr<rCr;r'r'r5num_quantizersszXcodecConfig.num_quantizers)__name__ __module__ __qualname____doc__ model_typerr sub_configsrlistfloatr?rr+rr)propertyr<r>r:rCr%rE __classcell__r'r'r3r5r sn2      Hr )rIr8typingrrnumpyr@ transformersrrrrconfiguration_utilsr utilsr get_loggerrFr/r __all__r'r'r'r5s