处理来自 s3 存储桶的音频文件以在 Python 中将语音转换为文本

Question

I have files stored in an s3 bucket that are uploaded from Genysis Purecloud. They are customer calls saved as.opus files but can be converted to.wav files when using a download function in python. I am having problems processing these files using the Python libraries boto3 and speech_recognition .

I need to be able to call the audio file (as variable f below) in the following script to firstly transform the audio into text and then run it through an NLP algorithm:

audio = f
r = sr.Recognizer()

with sr.AudioFile(audio) as source:
    # listen for the data (load audio to memory)
    audio_data = r.record(source)
    #recognize (convert from speech to text)
    text = r.recognize_google(audio_data)
    print(text)

I can use the get() function in boto3 to retrieve the object/file from s3:

s3_object = s3_resource.Bucket(bucket).Object('9d3be36b-6777-4b2a-8912-4dbfaaeea0ab/year=2022/month=11/day=10/hour=0/conversation_id=00d32818-16f5-4a6f-a97c-3db50ce9254f/36f1c0ce-d168-4df3-bae8-219bd5b59802.opus').get()

The problem is that the speech recognition library does not accept this as audio data as it is not reading the object as a file such as.wav.mp3 etc.

Any ideas on how I can pull the files from s3 so that they are in a format where I can process them? Thanks

I have tried using the get() function in boto3. I have also tried downloading the object onto my local drive. I managed to save this as a.wav file and when trying the process in the speech recognition library, get the follwing error:

'charmap' codec can't decode byte 0x90 in position 485: character maps to <undefined>

Answer 1

'charmap' codec can't decode byte 0x90 in position 485: character maps to

You opened the audio file in text mode. You need to open it in binary mode.

f = open('myfile.wav', 'rb')