Spaces:
Running
Running
| import wget | |
| import random | |
| import string | |
| import ffmpeg | |
| from .. import helpers | |
| import speech_recognition as sr | |
| def recognizeVoice(request): | |
| try: | |
| if request.method == 'POST': url = request.form['url'] | |
| else: url = request.args['url'] | |
| if url.strip() in ['', None]: raise Exception() | |
| except: return {"status": "error", "details": { "error_code": 101, "error_details": "No link provided" }} | |
| try: | |
| if request.method == 'POST': signature = request.form['signature'] | |
| else: signature = request.args['signature'] | |
| except: return {"status": "error", "details": { "error_code": 103, "error_details": "No signature" }} | |
| if not helpers.checkSignature(signature): return {"status": "error", "details": { "error_code": 105, "error_details": "Invalid signature" }} | |
| try: | |
| if request.method == 'POST': lang = request.form['lang'] | |
| else: lang = request.args['lang'] | |
| if lang.lower() in ['en','en-us']: lang = 'en-US' | |
| elif lang.lower() in ['ru','ru-ru']: lang = 'ru-RU' | |
| except: lang = "en-US" | |
| fileId = ''.join(random.SystemRandom().choice(string.ascii_uppercase + string.ascii_lowercase + string.digits) for _ in range(16)) | |
| fileExt = url[url.rfind('.'):url.rfind('.')+4] | |
| if fileExt in [".wav",".mp3",".ogg",'.aac']: pass | |
| else: return {"status": "error", "details": { "error_code": 111, "error_details": "Wrong file format (only ogg, wav, mp3)" }} | |
| r = sr.Recognizer() | |
| config = helpers.configFile() | |
| wget.download(url, f"{config['temp-path']}/{fileId}{fileExt}") | |
| if fileExt != ".wav": | |
| audio_input = ffmpeg.input(f"{config['temp-path']}/{fileId}{fileExt}") | |
| oldFE = fileExt | |
| fileExt = ".wav" | |
| audio_output = ffmpeg.output(audio_input.audio, f"{config['temp-path']}/{fileId}{fileExt}") | |
| ffmpeg.run(audio_output) | |
| helpers.deleteAudio(f"temp/{fileId}.{oldFE}") | |
| rawSource = sr.AudioFile(f"{config['temp-path']}/{fileId}{fileExt}") | |
| with rawSource as source: | |
| r.adjust_for_ambient_noise(source) | |
| audio = r.record(source) | |
| try: googleText = r.recognize_google(audio, language=lang) | |
| except: googleText = "" | |
| # at now here's no keys :( | |
| #try: houndifyText = r.recognize_houndify(audio) | |
| #except: houndifyText = "" | |
| helpers.deleteAudio(f"temp/{fileId}{fileExt}") | |
| return {"status": "pass", "result": {"google": googleText, "houndify": "NOT-WORKING"}} |