Add-Vishnu/Whisper_CPP_ASR_CLI

1import gradio as gr2import soundfile as sf3import tempfile4import shutil5import os6import librosa7import time8import numpy as np9import subprocess 10 11# command = r"""wine './whisper_blas_bin_v1_3_0/main.exe' -h"""12# wine_command = """sudo apt-get install wine"""13command2 = """chmod +777 ./whisper_blas_bin_v1_3_0/main.exe"""14# wine_c = subprocess.run(wine_command,shell=True,stdout=subprocess.PIPE,stderr=subprocess.PIPE, text=True)15perm = subprocess.run(command2, shell=True,stdout=subprocess.PIPE,stderr=subprocess.PIPE, text=True)16# result = subprocess.run(command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True)17# print("Wine Installation: ",wine_c)18print("Access Installation: ",perm)19# Fpr win32 instalattion while using medium model20# command3 = "apt install sudo"21# command4 = "dpkg --add-architecture i386"22# command5 = "apt-get update"23# command6 = "apt-get install wine32:i386"24# t1= subprocess.run(command3, shell=True,stdout=subprocess.PIPE,stderr=subprocess.PIPE, text=True)25# t2= subprocess.run(command4, shell=True,stdout=subprocess.PIPE,stderr=subprocess.PIPE, text=True)26# t3= subprocess.run(command5, shell=True,stdout=subprocess.PIPE,stderr=subprocess.PIPE, text=True)27# t4= subprocess.run(command6, shell=True,stdout=subprocess.PIPE,stderr=subprocess.PIPE, text=True)28# print("T1: ",t1)29# print("T2: ",t2)30# print("T3: ",t3)31# print("T4: ",t4)32 33 34def resample_to_16k(audio, orig_sr):35    y_resampled = librosa.resample(y=audio, orig_sr=orig_sr, target_sr = 16000)36    return y_resampled37 38def transcribe(audio):39    sr,y = audio40    y = y.astype(np.float32)41    y /= np.max(np.abs(y))42    y_resampled = resample_to_16k(y, sr)43    44    45    with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as temp_audio:46        temp_audio_path = temp_audio.name47        sf.write(temp_audio_path, y_resampled, 16000)48 49 50    51    # command = rf"""wine './whisper_blas_bin_v1_3_0/main.exe' -m './whisper_blas_bin_v1_3_0/models/ggml-model-whisper-small.en.bin' -osrt -f '{temp_audio_path}' -nt"""  # English only52    command = rf"""wine './whisper_blas_bin_v1_3_0/main.exe' -m './whisper_blas_bin_v1_3_0/models/ggml-model-whisper-base.bin' -osrt -f '{temp_audio_path}' -nt"""    # Multilingual53    # win32 error while using medium model54    # command = rf"""wine './whisper_blas_bin_v1_3_0/main.exe' -m './whisper_blas_bin_v1_3_0/models/ggml-model-whisper-medium-q5_0.bin' -osrt -f '{temp_audio_path}' -nt"""    # Multilingual55    56    start_time = time.time()57    result = subprocess.run(command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True)58    end_time = time.time()59    print("Output",result.stdout)60    print("Error",result.stderr)61    transcription = result.stdout62    print(transcription)63    64    print("--------------------------")65    print(f"Execution time: {end_time - start_time} seconds")66    return transcription, (end_time - start_time)67 68 69 70demo = gr.Interface(71    transcribe,72    inputs = "microphone",73    # gr.Audio(sources=["microphone"]),74    outputs = [gr.Textbox(label="CLI_Transcription"),gr.Textbox(label="Time taken for Transcription")],75    examples=["./Samples/Hindi_1.mp3","./Samples/Hindi_2.mp3","./Samples/Tamil_1.mp3","./Samples/Tamil_2.mp3","./Samples/Marathi_1.mp3","./Samples/Marathi_2.mp3","./Samples/Nepal_1.mp3","./Samples/Nepal_2.mp3","./Samples/Telugu_1.wav","./Samples/Telugu_2.wav","./Samples/Malayalam_1.wav","./Samples/Malayalam_2.wav","./Samples/Gujarati_1.wav","./Samples/Gujarati_2.wav","./Samples/Bengali_1.wav","./Samples/Bengali_2.wav"]76 77)78 79demo.launch()