Google Cloud Function: Audio Transcription

This repository contains a Python-based Google Cloud Function designed to automatically transcribe audio files stored in a Google Cloud Storage bucket. The transcription is processed using the Google Cloud Speech-to-Text API, and the resulting text file is saved in the same Cloud Storage bucket.

Features

• Transcribes audio files uploaded to a designated Google Cloud Storage bucket.
• Automatic encoding detection using Google Cloud Speech-to-Text API's AutoDetect feature.
• Creates a transcription text file and uploads it to the same bucket as the original audio file.

How It Works

1. Upload an audio file to the Cloud Storage bucket associated with the Google Cloud Function trigger.
2. The function is triggered by the finalized event (indicating a new file has been uploaded).
3. The function transcribes the audio using Google Cloud Speech-to-Text API.
4. The transcription is saved as a .txt file in the same Cloud Storage bucket.

Requirements

• Python Version: 3.9
• Google Cloud Services:
  • Google Cloud Functions
  • Google Cloud Storage
  • Google Cloud Speech-to-Text API

Setup and Deployment

1. Enable APIs

Ensure that the following APIs are enabled for your Google Cloud project:

• Cloud Functions API
• Cloud Storage API
• Cloud Speech-to-Text API

2. Deploy the Function Using the Google Cloud Console

Option 1: Using the Google Cloud Console (GUI)

As an alternative to using the command-line interface (CLI), you can deploy the function using the Google Cloud Console:

1. Go to the Cloud Functions section in the Google Cloud Console: Cloud Functions Console.
2. Click on Create Function.
3. In the GUI, specify the function name, region, and other settings.
4. Choose the trigger type as Cloud Storage and set the event to Finalize/Create.
5. For the source code, either:
   • Enter the source code manually in the inline editor.
   • Upload a ZIP file containing the source code.
6. Configure the entry point (e.g., speech_to_text), runtime (Python 3.9), and environment variables.
7. Click Deploy to launch your function.

Option 2: Using the CLI

Alternatively, you can deploy the function using the Google Cloud CLI:

gcloud functions deploy speech-to-text \
--runtime python39 \
--trigger-resource [BUCKET_NAME] \
--trigger-event google.storage.object.finalize \
--entry-point speech_to_text \
--region [REGION]

Replace:

• [BUCKET_NAME] with the name of your Cloud Storage bucket.
• [REGION] with the region where you want to deploy your function (e.g., us-central1).

3. Upload Audio Files

Once the function is deployed:

• Upload an audio file to the Cloud Storage bucket.
• The function will be triggered by the finalized event, transcribe the audio file, and write a .txt file with the transcription to the same bucket.

4. Verify the Output

• Check the Cloud Storage bucket for the newly created transcription file, which will have the same name as the original audio file but with a .txt extension.

Important Notes

• File Type Support: For a complete list of supported file types, refer to the Google Cloud Speech-to-Text documentation.

• Note: Although Google states it supports AAC, I was unable to transcribe AAC files in M4A and MP4 containers; it kept giving an InvalidArgument error.

• Note: Your Cloud Storage bucket region must match the region where your Cloud Function is deployed. This ensures proper communication and minimizes latency issues.