AIP-C01 · Question #8
A financial services company is developing a real-time generative AI (GenAI) assistant to support human call center agents. The GenAI assistant must transcribe live customer speech, analyze context…
The correct answer is B. Use Amazon Transcribe streaming with partial results enabled to deliver fragments of transcribed. Meeting sub-1-second end-to-end latency requires two streaming pipelines working together. Amazon Transcribe Streaming with partial results enabled sends transcribed text fragments to the application while the customer is still speaking, rather than waiting for a complete…
Question
A financial services company is developing a real-time generative AI (GenAI) assistant to support human call center agents. The GenAI assistant must transcribe live customer speech, analyze context, and provide incremental suggestions to call center agents while a customer is still speaking. To preserve responsiveness, the GenAI assistant must maintain end-to-end latency under 1 second from speech to initial response display. The architecture must use only managed AWS services and must support bidirectional streaming to ensure that call center agents receive updates in real time. Which solution will meet these requirements?
Options
- AUse Amazon Transcribe streaming to transcribe calls. Pass the text to Amazon Comprehend for
- BUse Amazon Transcribe streaming with partial results enabled to deliver fragments of transcribed
- CUse Amazon Transcribe batch processing to convert calls to text. Pass complete transcripts to
- DUse the Amazon Transcribe streaming API with an AWS Lambda function to transcribe each
How the community answered
(24 responses)- A13% (3)
- B79% (19)
- C4% (1)
- D4% (1)
Explanation
Meeting sub-1-second end-to-end latency requires two streaming pipelines working together. Amazon Transcribe Streaming with partial results enabled sends transcribed text fragments to the application while the customer is still speaking, rather than waiting for a complete sentence or utterance. This partial text is immediately forwarded to Amazon Bedrock's streaming inference API, which returns response tokens incrementally as they are generated. The combination achieves bidirectional streaming (speech in, tokens out) with AWS managed services. Option A uses Amazon Comprehend, which is not a generative AI service and does not provide incremental suggestions. Option C uses batch transcription, introducing seconds of delay per utterance, violating the <1 second requirement. Option D describes using Lambda to process each word individually, adding per-invocation cold start and execution overhead that breaks the latency budget.
Topics
Community Discussion
No community discussion yet for this question.