Build Your Own Audio Transcription Tool with AI

 In this video, you will learn how to deploy a LLM based application intro production by using Amazon Bedrock, Amazon Transcribe to summarize audio files with ASR model, Titan. 

Code:

import boto3

import json

import uuid

import time

from jinja2 import Template

bedrock_runtime = boto3.client('bedrock-runtime', region_name='us-east-1')

s3_client = boto3.client(service_name='s3', region_name='ap-southeast-2')

transcribe_client = boto3.client('transcribe', region_name='ap-southeast-2')

bucket_name='<replace your bucket name here>'

#file_name = 'angry.mp3'

file_name = 'happy.mp3'

job_name = 'transcription-job-' + str(uuid.uuid4())

response = transcribe_client.start_transcription_job(

    TranscriptionJobName=job_name,

    Media={'MediaFileUri': f's3://{bucket_name}/{file_name}'},

    MediaFormat='mp3',

    LanguageCode='en-US',

    OutputBucketName=bucket_name,

    Settings={

        'ShowSpeakerLabels': True,

        'MaxSpeakerLabels': 2

    }

)

while True:

    status = transcribe_client.get_transcription_job(TranscriptionJobName=job_name)

    if status['TranscriptionJob']['TranscriptionJobStatus'] in ['COMPLETED', 'FAILED']:

        break

    time.sleep(2)

print(status['TranscriptionJob']['TranscriptionJobStatus'])

if status['TranscriptionJob']['TranscriptionJobStatus'] == 'COMPLETED':

   

    # Load the transcript from S3.

    transcript_key = f"{job_name}.json"

    transcript_obj = s3_client.get_object(Bucket=bucket_name, Key=transcript_key)

    transcript_text = transcript_obj['Body'].read().decode('utf-8')

    transcript_json = json.loads(transcript_text)

   

    output_text = ""

    current_speaker = None

   

    items = transcript_json['results']['items']

   

    for item in items:

       

        speaker_label = item.get('speaker_label', None)

        content = item['alternatives'][0]['content']

       

        # Start the line with the speaker label:

        if speaker_label is not None and speaker_label != current_speaker:

            current_speaker = speaker_label

            output_text += f"\n{current_speaker}: "

           

        # Add the speech content:

        if item['type'] == 'punctuation':

            output_text = output_text.rstrip()

           

        output_text += f"{content} "

       

    # Save the transcript to a text file

    with open(f'{job_name}.txt', 'w') as f:

        f.write(output_text)

with open(f'{job_name}.txt', "r") as file:

    transcript = file.read()

template_string = """ I need to summarize a conversation. The transcript of the

conversation is between the <data> XML like tags.

<data>

{{transcript}}

</data>

The summary must contain a one word sentiment analysis, and

a list of issues, problems or causes of friction

during the conversation. The output must be provided in

JSON format shown in the following example.

Example output:

{

    "sentiment": <sentiment>,

    "issues": [

        {

            "topic": <topic>,

            "summary": <issue_summary>,

        }

    ]

}

Write the JSON output and nothing more.

Here is the JSON output:   """

data = {

    'transcript' : transcript

}

template = Template(template_string)

prompt = template.render(data)

kwargs = {

    "modelId": "amazon.titan-text-express-v1",

    "contentType": "application/json",

    "accept": "*/*",

    "body": json.dumps(

        {

            "inputText": prompt,

            "textGenerationConfig": {

                "maxTokenCount": 512,

                "temperature": 0,

                "topP": 0.9

            }

        }

    )

}

response = bedrock_runtime.invoke_model(**kwargs)

response_body = json.loads(response.get('body').read())

generation = response_body['results'][0]['outputText']

print(generation)