Spaces:
Runtime error
Runtime error
| import streamlit as st | |
| import os | |
| os.system("pip install git+https://github.com/openai/whisper.git") | |
| import whisper | |
| from whisper import utils | |
| import ffmpeg | |
| import os | |
| from transformers import pipeline | |
| from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq | |
| import numpy as np | |
| SAMPLE_RATE = 16000 | |
| def load_audio(file: str, sr: int = SAMPLE_RATE): | |
| # This launches a subprocess to decode audio while down-mixing and resampling as necessary. | |
| # Requires the ffmpeg CLI and `ffmpeg-python` package to be installed. | |
| out, _ = ( | |
| ffmpeg.input(file, threads=0) | |
| .output("-", format="s16le", acodec="pcm_s16le", ac=1, ar=sr) | |
| .run(cmd=["ffmpeg", "-nostdin"], capture_stdout=True, capture_stderr=True) | |
| ) | |
| return np.frombuffer(out, np.int16).flatten().astype(np.float32) / 32768.0 | |
| if 'processor' not in locals(): | |
| with st.spinner('Wait for it...'): | |
| # processor = AutoProcessor.from_pretrained("openai/whisper-tiny") | |
| model=whisper.load_model("tiny") | |
| #pipe = pipeline('sentiment-analysis') | |
| #pipe2 = pipeline(task="image-bestify", model="beihai/GFPGAN-V1.3-whole-image") | |
| #text = st.text_area('entre com algum texto') | |
| #st.title("Wav a ser transcrito ") | |
| wav_up = st.file_uploader("Upload",type=['wav','ogg','mp3']) | |
| if wav_up is not None: | |
| file_details = {"FileName":wav_up.name,"FileType":wav_up.type} | |
| st.write(file_details) | |
| with open(wav_up.name,"wb") as f: | |
| f.write(wav_up.getbuffer()) | |
| st.success("Saved File") | |
| audio = whisper.load_audio(wav_up.name) | |
| audio = whisper.pad_or_trim(audio) | |
| st.audio(wav_up.name, format="audio/wav", start_time=0) | |
| if st.button('Processa'): | |
| if wav_up is not None: | |
| with st.spinner('Wait for it...'): | |
| # print('2') | |
| # transcription=model.transcribe( | |
| # audio, | |
| # language = None | |
| # ) | |
| #processado=np.frombuffer(wav_up.getbuffer(), np.int16).flatten().astype(np.float32) / 32768.0 | |
| # input_features = processor(audio , return_tensors="pt").input_features | |
| # forced_decoder_ids = processor.get_decoder_prompt_ids(language = None, task = "transcribe") | |
| transcription=model.transcribe( | |
| audio, | |
| language = 'pt' | |
| ) | |
| # predicted_ids = model.generate(input_features, forced_decoder_ids = forced_decoder_ids) | |
| #transcription = processor.batch_decode(predicted_ids, skip_special_tokens = True) | |
| string1='' | |
| for i, segment in enumerate(transcription, start=1): | |
| # write srt lines | |
| string1=string1+"\n" + f"{str(i)}\n"+ f"{utils.format_timestamp(segment['start'], always_include_hours=True, decimal_marker=',')} --> " + f"{utils.format_timestamp(segment['end'], always_include_hours=True, decimal_marker=',')}\n"+ f"{segment['text'].strip().replace('-->', '->')}\n"; | |
| st.write(string1 ) | |
| st.success("Texto Gerado") | |
| # print('3') | |