devilent2/whisper-v3-zero-canary
1
1o
2 �f� � 3 @ s@ d dl Z d dlZd dlmZ d dlZd dlZd dlZd dl m4Z5 d dl m Z dZ
dZe j�� r1d ndZedkr9dZ
dd � Zee
�aejd6d � �Zd d lmZ d
d� Zdd� Zdd� Zdd� Zdd� Z e� � Z ej eej dd�ej!d dgdd d�ej"d e
d d d �gej#d d!�ej#d"d!�gd#d$d%d&d'�Z$ej eej dd(d)�ej!d dgdd d�ej"d e
d d*d �gej#d d!�ej#d"d!�gd#d$d%d&d'�Z%e �N e�&e$e%gd+d(g� e�'� �
e�(d,�Z)W d � n1 s�w Y ej"d-d.d/�Z*ej"d0d.d/�Z+e)�,e de*e+g� e j-ede*e+gd1d.d2� W d � n 1 �sw Y e j.d3d4� dS )5� N)�pipeline)� ffmpeg_readz distil-whisper/distil-large-v3� �cpuzopenai/whisper-tinyc C s t d| dtd�S )Nz automatic-speech-recognition� )�task�model�chunk_length_s�device)r r7 )�8model_name� r �5/teamspace/studios/this_studio/whisper-v3-zero/app.py�
load_pipeline s �r c 9 C s� | d u r t �d��|tjjkrt|�at�� }t�| �}|j }t| t10d|idd�d }t�� }|| }d|d�d|d�d |� d11t j �
� rFd nd � �} || fS )
Nz^No audio file submitted! Please upload or record an audio file before submitting your request.r T)�12batch_size�generate_kwargs�return_timestamps�textzTranscription Time: �.2fz seconds13Audio Duration: z seconds14Model Used: z15Device Used: �GPU�CPU)�gr�Error�piper � name_or_pathr �time�mp�
AudioFileClip�duration�16BATCH_SIZE�torch�cuda� is_available)17�inputsr r �18start_time�audio�audio_durationr �end_time�transcription_time�transcription_time_outputr r r
�19transcribe s& 20 2122����r) )�GPUStatCollectionc 23 C s� t j�� dkr dS z;t�� } | D ]1}|j}|j}|j}|j}|j }|| d }d|� d|� d|� d|� d|� d |d24�d �
}| W S W d S t25y` } zt d | � �� t � W Y d } ~ S d } ~ ww )
NFzNo Nviadia Device�d zGPU z: z, Utilization: z%, Memory Used: zMB, Memory Total: zMB, Memory Utilization: r �%zError getting GPU stats: )
r r r! r* � new_query�index�name� utilization� memory_used� memory_total� Exception�print�torch_update_gpu_status)26� gpu_stats�gpu�gpu_id�gpu_name�gpu_utilizationr1 r2 �memory_utilization�27gpu_status�er r r
�update_gpu_status; s& ,28� ��r> c C sj t j�� r1t j�d�} t j�d�}|d d }|d |d d }d| � d|d�d|d�d�}|S d }|S )29Nr � i zGPU: z30Total Memory: r z MB31Used Memory: z MBzNo GPU available)r r r! �get_device_name� mem_get_info)�gpu_infoZ32gpu_memory� total_memoryZ used_memoryr< r r r
r5 O s 33 �r5 c C s< dd l } | j �� �� }|�d�}t�� }d|� d|� �}|S )Nr z%H:%M:%Sz CPU Usage: z% )�datetime�nowr �strftime�psutil� cpu_percent)rD � current_time�time_strrH �34cpu_statusr r r
�update_cpu_status[ s 35rL c C s t � } t� }| |fS �N)r> rL )r< rK r r r
�
update_statusf s rN c C s t � S rM )rN r r r r
�refresh_statusk s rO �filepath)�type� translate�Task)�label�valuez36Model NamezEnter the model namea Some available models: distil-whisper/distil-large-v3 distil-whisper/distil-medium.en Systran/faster-distil-whisper-large-v3 Systran/faster-whisper-large-v3 Systran/faster-whisper-medium openai/whisper-tiny, openai/whisper-base, openai/whisper-medium, openai/whisper-large-v3)rT rU � placeholder�info�
Transcription)rT zTranscription Info� huggingfacezWhisper Transcriptionu� Transcribe long-form microphone or audio inputs with the click of a button! Demo uses the specified OpenAI Whisper checkpoint and 🤗 Transformers to transcribe audio files of arbitrary length.�never)�fnr" �outputs�theme�title� description�allow_flaggingz37Audio file)rQ rT zoSome available models: openai/whisper-tiny, openai/whisper-base, openai/whisper-medium, openai/whisper-large-v2�38MicrophonezRefresh Statusz39GPU StatusF)rT � interactivez40CPU Status� )r" r\ �every�queueT)�share)/r r �moviepy.editor�editorr rG �gradior �spaces� transformersr �"transformers.pipelines.audio_utilsr �DEFAULT_MODEL_NAMEr r r! r41 r r r r) �gpustatr* r> r5 rL rN rO �Blocks�demo� Interface�Audio�Radio�Textbox�TextArea�
mf_transcribe�file_transcribe�TabbedInterface�Row�Button�refresh_button�gpu_status_output�cpu_status_output�click�load�launchr r r r
�<module> s� 42 43��44� ��45�46 � �