This Lambda is invoked by another Lambda.
Get the audio stream from KVS.
Get the real time transcription by Google Speech to Text.
Write the transcription to DynamoDB.
Save the audio data as wav file in s3.

https://drive.google.com/file/d/1OsolNuuSxA6zbFeaM-fIVdolntdGaLqo/view?usp=sharing

• openjdk version "1.8.0_265"
• Gradle 6.6.1
• credentials.json with Google Speech To Text enabled

Maybe it will work with other versions as well. Try it for yourself.

git clone https://github.com/yuga-t/kvsConsumerGoogleSTT.git

cd kvsConsumerGoogleSTT

Put credentials.json in kvsConsumerGoogleSTT/src/resources/

gradle build

Upload kvsConsumerGoogleSTT/build/ditributions/kvsConsumerGoogleSTT.zip to Lambda.

Set Handler of Runtime settings

com.amazonaws.kvstranscribestreaming.KVSTranscribeStreamingLambda::handleRequest

Set Environment Variable of Lambda.

• key:GOOGLE_APPLICATION_CREDENTIALS, value:credentials.json
• key:LIMITDAY_OF_TTL, value:1
• key:RECORDINGS_BUCKET_NAME, value:amazon-connect-audio-bucket
• key:RECORDINGS_KEY_PREFIX, value:recordings/
• key:RECORDINGS_PUBLIC_READ_ACL, value:TRUE
• key:REGION, value:ap-northeast-1
• key:START_SELECTOR_TYPE, value:NOW
• key:TRANSCRIPT_TABLE_NAME, value:transcriptTable

The demo will not work on this project alone.
Requires AmazonConnect ContactFlow and another Lambda, and so on.
Please refer to here.

• https://github.com/amazon-connect/amazon-connect-realtime-transcription
• https://github.com/googleapis/java-speech/blob/master/samples/snippets/src/main/java/com/example/speech/InfiniteStreamRecognize.java

"kvsConsumerGoogleSTT" is under MIT license.