Machine Learning
Amazon Transcribe
Amazon Transcribe converts streaming or batch audio into text with timestamps, channels, speakers, and supported language features. It prepares speech for summarization, retrieval, and agent-assistance pipelines.
Key points
- Streaming transcription returns partial and final results
- Custom vocabularies improve recognition of domain terms
- PII identification or redaction is available only for supported modes and languages
When to use it
- Transcribe contact-center audio before summarization
- Create timestamped searchable transcripts for a media knowledge base
Exam tips
- Choose Transcribe for speech-to-text and Amazon Lex for intent-driven conversational bots
- Wait for final segments before durable downstream actions and verify language-specific feature support