# Inspiration from https://huggingface.co/spaces/vumichien/whisper-speaker-diarization import whisper import datetime import subprocess import gradio as gr from pathlib import Path import pandas as pd import re import time import os import numpy as np from sklearn.cluster import AgglomerativeClustering from pytube import YouTube import torch import pyannote.audio from pyannote.audio.pipelines.speaker_verification import PretrainedSpeakerEmbedding from pyannote.audio import Audio from pyannote.core import Segment from gpuinfo import GPUInfo import wave import contextlib from transformers import pipeline import psutil from zipfile import ZipFile from io import StringIO import csv # ---- Model Loading ---- whisper_models = ["base", "small", "medium", "large"] source_languages = { "en": "English", "de": "German", "es": "Spanish", "fr": "French", } source_language_list = [key[0] for key in source_languages.items()] MODEL_NAME = "openai/whisper-small" lang = "en" device = "cuda" if torch.cuda.is_available() else "cpu" pipe = pipeline( task="automatic-speech-recognition", model=MODEL_NAME, chunk_length_s=30, device=device, ) pipe.model.config.forced_decoder_ids = pipe.tokenizer.get_decoder_prompt_ids(language=lang, task="transcribe") embedding_model = PretrainedSpeakerEmbedding( "speechbrain/spkrec-ecapa-voxceleb", device=torch.device("cuda" if torch.cuda.is_available() else "cpu")) # ---- S2T & Speaker diarization ---- def transcribe(microphone, file_upload): warn_output = "" if (microphone is not None) and (file_upload is not None): warn_output = ( "WARNING: You've uploaded an audio file and used the microphone. " "The recorded file from the microphone will be used and the uploaded audio will be discarded.\n" ) elif (microphone is None) and (file_upload is None): return "ERROR: You have to either use the microphone or upload an audio file" file = microphone if microphone is not None else file_upload text = pipe(file)["text"] return warn_output + text def convert_time(secs): return datetime.timedelta(seconds=round(secs)) def convert_to_wav(filepath): _,file_ending = os.path.splitext(f'{filepath}') audio_file = filepath.replace(file_ending, ".wav") print("starting conversion to wav") os.system(f'ffmpeg -i "{filepath}" -ar 16000 -ac 1 -c:a pcm_s16le "{audio_file}"') return audio_file def speech_to_text(microphone, file_upload, selected_source_lang, whisper_model, num_speakers): """ # Transcribe audio file and separate into segment, assign speakers to segments 1. Using Open AI's Whisper model to seperate audio into segments and generate transcripts. 2. Generating speaker embeddings for each segments. 3. Applying agglomerative clustering on the embeddings to identify the speaker for each segment. Speech Recognition is based on models from OpenAI Whisper https://github.com/openai/whisper Speaker diarization model and pipeline from by https://github.com/pyannote/pyannote-audio """ model = whisper.load_model(whisper_model) time_start = time.time() try: # Read and convert audio file warn_output = "" if (microphone is not None) and (file_upload is not None): warn_output = ( "WARNING: You've uploaded an audio file and used the microphone. " "The recorded file from the microphone will be used and the uploaded audio will be discarded.\n" ) elif (microphone is None) and (file_upload is None): return "ERROR: You have to either use the microphone or upload an audio file" file = microphone if microphone is not None else file_upload if microphone is None and file_upload is not None: file = convert_to_wav(file) # Get duration with contextlib.closing(wave.open(file,'r')) as f: frames = f.getnframes() rate = f.getframerate() duration = frames / float(rate) print(f"conversion to wav ready, duration of audio file: {duration}") # Transcribe audio options = dict(language=selected_source_lang, beam_size=3, best_of=3) transcribe_options = dict(task="transcribe", **options) result = model.transcribe(file, **transcribe_options) segments = result["segments"] print("whisper done with transcription") except Exception as e: raise RuntimeError("Error converting audio file") try: # Create embedding def segment_embedding(segment): audio = Audio() start = segment["start"] # Whisper overshoots the end timestamp in the last segment end = min(duration, segment["end"]) clip = Segment(start, end) waveform, sample_rate = audio.crop(file, clip) return embedding_model(waveform[None]) embeddings = np.zeros(shape=(len(segments), 192)) for i, segment in enumerate(segments): embeddings[i] = segment_embedding(segment) embeddings = np.nan_to_num(embeddings) print(f'Embedding shape: {embeddings.shape}') # Assign speaker label if num_speakers == 1: for i in range(len(segments)): segments[i]["speaker"] = 'SPEAKER 1' else: clustering = AgglomerativeClustering(num_speakers).fit(embeddings) labels = clustering.labels_ for i in range(len(segments)): segments[i]["speaker"] = 'SPEAKER ' + str(labels[i] + 1) # Make output objects = { 'Start' : [], 'End': [], 'Speaker': [], 'Text': [] } text = '' if num_speakers == 1: objects['Start'].append(str(convert_time(segment["start"]))) objects['Speaker'].append(segment["speaker"]) for (i, segment) in enumerate(segments): text += segment["text"] + ' ' objects['Text'].append(text) objects['End'].append(str(convert_time(segment["end"]))) else: for (i, segment) in enumerate(segments): if i == 0 or segments[i - 1]["speaker"] != segment["speaker"]: objects['Start'].append(str(convert_time(segment["start"]))) objects['Speaker'].append(segment["speaker"]) if i != 0: objects['End'].append(str(convert_time(segments[i - 1]["end"]))) objects['Text'].append(text) text = '' text += segment["text"] + ' ' objects['End'].append(str(convert_time(segments[i - 1]["end"]))) objects['Text'].append(text) time_end = time.time() time_diff = time_end - time_start memory = psutil.virtual_memory() gpu_utilization, gpu_memory = GPUInfo.gpu_usage() gpu_utilization = gpu_utilization[0] if len(gpu_utilization) > 0 else 0 gpu_memory = gpu_memory[0] if len(gpu_memory) > 0 else 0 system_info = f""" *Memory: {memory.total / (1024 * 1024 * 1024):.2f}GB, used: {memory.percent}%, available: {memory.available / (1024 * 1024 * 1024):.2f}GB.* *Processing time: {time_diff:.5} seconds.* *GPU Utilization: {gpu_utilization}%, GPU Memory: {gpu_memory}MiB.* """ return pd.DataFrame(objects), system_info except Exception as e: raise RuntimeError("Error Running inference with local model", e) # ---- Youtube Conversion ---- def get_youtube(video_url): yt = YouTube(video_url) abs_video_path = yt.streams.filter(progressive=True, file_extension='mp4').order_by('resolution').desc().first().download() print("Success download video") print(abs_video_path) return abs_video_path def yt_to_text(video_file_path, selected_source_lang, whisper_model, num_speakers): """ # Transcribe youtube link using OpenAI Whisper 1. Using Open AI's Whisper model to seperate audio into segments and generate transcripts. 2. Generating speaker embeddings for each segments. 3. Applying agglomerative clustering on the embeddings to identify the speaker for each segment. Speech Recognition is based on models from OpenAI Whisper https://github.com/openai/whisper Speaker diarization model and pipeline from by https://github.com/pyannote/pyannote-audio """ model = whisper.load_model(whisper_model) time_start = time.time() if(video_file_path == None): raise ValueError("Error no video input") print(video_file_path) try: # Read and convert youtube video _,file_ending = os.path.splitext(f'{video_file_path}') print(f'file ending is {file_ending}') audio_file = video_file_path.replace(file_ending, ".wav") print("starting conversion to wav") os.system(f'ffmpeg -i "{video_file_path}" -ar 16000 -ac 1 -c:a pcm_s16le "{audio_file}"') # Get duration with contextlib.closing(wave.open(audio_file,'r')) as f: frames = f.getnframes() rate = f.getframerate() duration = frames / float(rate) print(f"conversion to wav ready, duration of audio file: {duration}") # Transcribe audio options = dict(language=selected_source_lang, beam_size=5, best_of=5) transcribe_options = dict(task="transcribe", **options) result = model.transcribe(audio_file, **transcribe_options) segments = result["segments"] print("starting whisper done with whisper") except Exception as e: raise RuntimeError("Error converting video to audio") try: # Create embedding def segment_embedding(segment): audio = Audio() start = segment["start"] # Whisper overshoots the end timestamp in the last segment end = min(duration, segment["end"]) clip = Segment(start, end) waveform, sample_rate = audio.crop(audio_file, clip) return embedding_model(waveform[None]) embeddings = np.zeros(shape=(len(segments), 192)) for i, segment in enumerate(segments): embeddings[i] = segment_embedding(segment) embeddings = np.nan_to_num(embeddings) print(f'Embedding shape: {embeddings.shape}') # Assign speaker label if num_speakers == 1: for i in range(len(segments)): segments[i]["speaker"] = 'SPEAKER 1' else: clustering = AgglomerativeClustering(num_speakers).fit(embeddings) labels = clustering.labels_ for i in range(len(segments)): segments[i]["speaker"] = 'SPEAKER ' + str(labels[i] + 1) # Make output objects = { 'Start' : [], 'End': [], 'Speaker': [], 'Text': [] } text = '' if num_speakers == 1: objects['Start'].append(str(convert_time(segment["start"]))) objects['Speaker'].append(segment["speaker"]) for (i, segment) in enumerate(segments): text += segment["text"] + ' ' objects['Text'].append(text) objects['End'].append(str(convert_time(segment["end"]))) else: for (i, segment) in enumerate(segments): if i == 0 or segments[i - 1]["speaker"] != segment["speaker"]: objects['Start'].append(str(convert_time(segment["start"]))) objects['Speaker'].append(segment["speaker"]) if i != 0: objects['End'].append(str(convert_time(segments[i - 1]["end"]))) objects['Text'].append(text) text = '' text += segment["text"] + ' ' objects['End'].append(str(convert_time(segments[i - 1]["end"]))) objects['Text'].append(text) time_end = time.time() time_diff = time_end - time_start memory = psutil.virtual_memory() gpu_utilization, gpu_memory = GPUInfo.gpu_usage() gpu_utilization = gpu_utilization[0] if len(gpu_utilization) > 0 else 0 gpu_memory = gpu_memory[0] if len(gpu_memory) > 0 else 0 system_info = f""" *Memory: {memory.total / (1024 * 1024 * 1024):.2f}GB, used: {memory.percent}%, available: {memory.available / (1024 * 1024 * 1024):.2f}GB.* *Processing time: {time_diff:.5} seconds.* *GPU Utilization: {gpu_utilization}%, GPU Memory: {gpu_memory}MiB.* """ return pd.DataFrame(objects), system_info except Exception as e: raise RuntimeError("Error Running inference with local model", e) def download_csv(dataframe: pd.DataFrame): compression_options = dict(method='zip', archive_name='output.csv') dataframe.to_csv('output.zip', index=False, compression=compression_options) return 'output.zip' # ---- Gradio Layout ---- # Inspiration from https://huggingface.co/spaces/vumichien/whisper-speaker-diarization # -- General Functions -- df_init = pd.DataFrame(columns=['Start', 'End', 'Speaker', 'Text']) memory = psutil.virtual_memory() title = "Whisper speaker diarization & speech recognition" interface = gr.Blocks(title=title) interface.encrypt = False # -- Functions Audio Input -- microphone_in = gr.inputs.Audio(source="microphone", type="filepath", optional=True) upload_in = gr.inputs.Audio(source="upload", type="filepath", optional=True) selected_source_lang_audio = gr.Dropdown(choices=source_language_list, type="value", value="en", label="Spoken language in audio", interactive=True) selected_whisper_model_audio = gr.Dropdown(choices=whisper_models, type="value", value="base", label="Selected Whisper model", interactive=True) number_speakers_audio = gr.Number(precision=0, value=2, label="Selected number of speakers", interactive=True) system_info_audio = gr.Markdown(f"*Memory: {memory.total / (1024 * 1024 * 1024):.2f}GB, used: {memory.percent}%, available: {memory.available / (1024 * 1024 * 1024):.2f}GB*") transcription_df_audio = gr.DataFrame(value=df_init, label="Transcription dataframe", row_count=(0, "dynamic"), max_rows = 10, wrap=True, overflow_row_behaviour='paginate') csv_download_audio = gr.outputs.File(label="Download CSV") # -- Functions Video Input -- video_in = gr.Video(label="Video file", mirror_webcam=False) youtube_url_in = gr.Textbox(label="Youtube url", lines=1, interactive=True) selected_source_lang_yt = gr.Dropdown(choices=source_language_list, type="value", value="en", label="Spoken language in audio", interactive=True) selected_whisper_model_yt = gr.Dropdown(choices=whisper_models, type="value", value="base", label="Selected Whisper model", interactive=True) number_speakers_yt = gr.Number(precision=0, value=2, label="Selected number of speakers", interactive=True) system_info_yt = gr.Markdown(f"*Memory: {memory.total / (1024 * 1024 * 1024):.2f}GB, used: {memory.percent}%, available: {memory.available / (1024 * 1024 * 1024):.2f}GB*") transcription_df_yt = gr.DataFrame(value=df_init, label="Transcription dataframe", row_count=(0, "dynamic"), max_rows = 10, wrap=True, overflow_row_behaviour='paginate') csv_download_yt = gr.outputs.File(label="Download CSV") with interface: with gr.Tab("Whisper speaker diarization & speech recognition"): gr.Markdown('''