Transcribing a Local Audio File using AssemblyAI

Transcribing mp3 files

Before building the UI, we need a few helper functions which we can use to upload the file to AssemblyAI’s server for the model to work and return the transcribed text.

The helper functions code is written in the transcribe.py file.

Import the required modules

The following piece of code should be present at the beginning of the transcribe.py file.

import os
from dotenv import load_dotenv
import requests

Helper function 1: Uploading a local audio file to AssemblyAI

The first function we need to write is a method to upload an audio file. This function should be present inside the transcribe.py file.

The AssemblyAI model expects the file to be accessible via a URL. Therefore, we’ll need to upload the audio file to blob storage to make it accessible via a URL. Fortunately, AssemblyAI provides a quick and easy way to do this.

We need to make a POST request to the following AssemblyAI API endpoint: https://api.assemblyai.com/v2/upload

The response will contain a temporary URL to the file, we can pass this URL back to the AssemblyAI transcript API endpoint. The URL is a private URL accessible only to the AssemblyAI servers.

All the uploads are immediately deleted after transcription and never stored.

We will use Python’s request library that we installed earlier to make the POST request.

def get_url(token,data):
    '''
    Parameter:
    token: The API key
    data : The File Object to upload
    Return Value:
    url : Url to uploaded file
    '''
    headers = {‘authorization’: token}
    response = requests.post(‘https://api.assemblyai.com/v2/upload',
    headers=headers,
    data=data)
    url = response.json()[“upload_url”]
    print(“Uploaded File and got temporary URL to file”)
    return URL

Explanation

• The function accepts the API token and the file object as parameters

• We make a POST request to the AssemblyAI Upload API endpoint mentioned above and include the API token and the file object as a part of the request body.

• The response object contains the URL to the uploaded file. This URL is returned by the function.