Transcribing multi-channel audio - Amazon Transcribe
Using channel identification in a batch transcriptionUsing channel identification in a streaming transcription

Transcribing multi-channel audio

If your audio has two channels, you can use channel identification to transcribe the speech from each channel separately. Amazon Transcribe doesn't currently support audio with more than two channels.

In your transcript, channels are assigned the labels ch_0 and ch_1.

In addition to the standard transcript sections (transcripts and items), requests with channel identification enabled include a channel_labels section. This section contains each utterance or punctuation mark, grouped by channel, and its associated channel label, timestamps, and confidence score.

"channel_labels": {
    "channels": [
        {
            "channel_label": "ch_0",
            "items": [                                      
                {
                    "channel_label": "ch_0",
                    "start_time": "4.86",
                    "end_time": "5.01",
                    "alternatives": [
                        {
                            "confidence": "1.0",
                            "content": "I've"
                        }
                    ],
                    "type": "pronunciation"
                },
                ...               
            "channel_label": "ch_1",
            "items": [
                {
                    "channel_label": "ch_1",
                    "start_time": "8.5",
                    "end_time": "8.89",
                    "alternatives": [
                        {
                            "confidence": "1.0",
                            "content": "Sorry"
                        }
                    ],
                    "type": "pronunciation"
                },
                ...
            "number_of_channels": 2
        },

Note that if a person on one channel speaks at the same time as a person on a separate channel, timestamps for each channel overlap while the individuals are speaking over each other.

To view a complete example transcript with channel identification, see Example channel identification output (batch).

Using channel identification in a batch transcription

To identify channels in a batch transcription, you can use the AWS Management Console, AWS CLI, or AWS SDKs; see the following for examples:

  1. Sign in to the AWS Management Console.

  2. In the navigation pane, choose Transcription jobs, then select Create job (top right). This opens the Specify job details page.

    Amazon Transcribe console 'Specify job details' page. In the 'Job settings' panel, you can specify a name for your transcription job, select a Model type, and specify your language settings.

  3. Fill in any fields you want to include on the Specify job details page, then select Next. This takes you to the Configure job - optional page.

    In the Audio settings panel, select Channel identification (under the 'Audio identification type' heading).

    Amazon Transcribe console 'Configure job' page. In the 'Audio settings' panel, you can enable Channel identification.

  4. Select Create job to run your transcription job.

This example uses the start-transcription-job. For more information, see StartTranscriptionJob.

aws transcribe start-transcription-job \
--region us-west-2 \
--transcription-job-name my-first-transcription-job \
--media MediaFileUri=s3://DOC-EXAMPLE-BUCKET/my-input-files/my-media-file.flac \
--output-bucket-name DOC-EXAMPLE-BUCKET \
--output-key my-output-files/ \
--language-code en-US \
--settings ChannelIdentification=true

Here's another example using the start-transcription-job command, and a request body that enables channel identification with that job.

aws transcribe start-transcription-job \
--region us-west-2 \
--cli-input-json file://my-first-transcription-job.json

The file my-first-transcription-job.json contains the following request body.

{
    "TranscriptionJobName": "my-first-transcription-job",
    "Media": {
        "MediaFileUri": "s3://DOC-EXAMPLE-BUCKET/my-input-files/my-media-file.flac"
  },
    "OutputBucketName": "DOC-EXAMPLE-BUCKET",
    "OutputKey": "my-output-files/", 
    "LanguageCode": "en-US",
    "Settings": {
        "ChannelIdentification": true
    }
}

This example uses the AWS SDK for Python (Boto3) to identify channels using the start_transcription_job method. For more information, see StartTranscriptionJob.

from __future__ import print_function
import time
import boto3
transcribe = boto3.client('transcribe', 'us-west-2')
job_name = "my-first-transcription-job"
job_uri = "s3://DOC-EXAMPLE-BUCKET/my-input-files/my-media-file.flac"
transcribe.start_transcription_job(
    TranscriptionJobName = job_name,
    Media = {
        'MediaFileUri': job_uri
    },
    OutputBucketName = 'DOC-EXAMPLE-BUCKET',
    OutputKey = 'my-output-files/', 
    LanguageCode = 'en-US', 
    Settings = {
        'ChannelIdentification':True
    }
)

while True:
    status = transcribe.get_transcription_job(TranscriptionJobName = job_name)
    if status['TranscriptionJob']['TranscriptionJobStatus'] in ['COMPLETED', 'FAILED']:
        break
    print("Not ready yet...")
    time.sleep(5)
print(status)

Using channel identification in a streaming transcription

To identify channels in a streaming transcription, you can use HTTP/2 or WebSockets; see the following for examples:

This example creates an HTTP/2 request that separates channels in your transcription output. For more information on using HTTP/2 streaming with Amazon Transcribe, see Setting up an HTTP/2 stream. For more detail on parameters and headers specific to Amazon Transcribe, see StartStreamTranscription.

POST /stream-transcription HTTP/2
host: transcribestreaming.us-west-2.amazonaws.com
X-Amz-Target: com.amazonaws.transcribe.Transcribe.StartStreamTranscription
Content-Type: application/vnd.amazon.eventstream
X-Amz-Content-Sha256: string
X-Amz-Date: 20220208T235959Z
Authorization: AWS4-HMAC-SHA256 Credential=access-key/20220208/us-west-2/transcribe/aws4_request, SignedHeaders=content-type;host;x-amz-content-sha256;x-amz-date;x-amz-target;x-amz-security-token, Signature=string
x-amzn-transcribe-language-code: en-US
x-amzn-transcribe-media-encoding: flac
x-amzn-transcribe-sample-rate: 16000      
x-amzn-channel-identification: TRUE
transfer-encoding: chunked

Parameter definitions can be found in the API Reference; parameters common to all AWS API operations are listed in the Common Parameters section.

This example creates a presigned URL that separates channels in your transcription output. Line breaks have been added for readability. For more information on using WebSocket streams with Amazon Transcribe, see Setting up a WebSocket stream. For more detail on parameters, see StartStreamTranscription.

GET wss://transcribestreaming.us-west-2.amazonaws.com:8443/stream-transcription-websocket?
&X-Amz-Algorithm=AWS4-HMAC-SHA256
&X-Amz-Credential=AKIAIOSFODNN7EXAMPLE%2F20220208%2Fus-west-2%2Ftranscribe%2Faws4_request
&X-Amz-Date=20220208T235959Z
&X-Amz-Expires=300
&X-Amz-Security-Token=security-token
&X-Amz-Signature=string
&X-Amz-SignedHeaders=content-type%3Bhost%3Bx-amz-date
&language-code=en-US
&specialty=PRIMARYCARE
&type=DICTATION
&media-encoding=flac
&sample-rate=16000    
&channel-identification=TRUE

Parameter definitions can be found in the API Reference; parameters common to all AWS API operations are listed in the Common Parameters section.