Amazon Transcribe streaming output repeats partial results - how to print only the final transcript

TL;DR

The fix is in the event handler, not the API call. That gives you one final line per utterance instead of the streaming repeats.

Steps

1. Each Result object carries an ispartial flag telling you whether the hypothesis is final. Only emit results where result.ispartial is False. The maintainer's example handler: class MyEventHandler(TranscriptResultStreamHandler): async def handletranscriptevent(self, transcriptevent): results = transcriptevent.transcript.results; for result in results: if result.is_partial is False: for alt in result.alternatives: ... handle alt.transcript here. That gives you one final line per utterance instead of the streaming repeats.

When to use

You are seeing this: A user streaming audio through Amazon Transcribe got output with every partial hypothesis repeated: This. / This is / This is a / This is a te / This is a test. Use this skill when you run into "Amazon Transcribe streaming output repeats partial results - how to print only the final transcript".

When not to use

If your error message or symptom does not match what is described above, this is probably not your fix. Search for your exact error text instead of forcing this one to fit.

Versions

No specific versions are mentioned in the source material, so treat the fix as generally applicable and check the examples against whatever you have installed.

Why this happens

A user streaming audio through Amazon Transcribe got output with every partial hypothesis repeated: This. / This is / This is a / This is a te / This is a test.

In the thread Amazon Transcribe streaming output repeats partial results - how to print only the final transcript, the question was: A user streaming audio through Amazon Transcribe got output with every partial hypothesis repeated: This. / This is / This is a / This is a te / This is a test. They only wanted the final transcript line. The examples print every result the service returns, and Transcribe emits partial hypotheses continuously as it refines them. The fix is in the event handler, not the API call. The verified resolution: Each Result object carries an ispartial flag telling you whether the hypothesis is final. Only emit results where result.ispartial is False. The maintainer's example handler: class MyEventHandler(TranscriptResultStreamHandler): async def handletranscriptevent(self, transcriptevent): results = transcriptevent.transcript.results; for result in results: if result.ispartial is False: for alt in result.alternatives: ... handle alt.transcript here. That gives you one final line per utterance instead of the streaming repeats. Original discussion: https://vectle.com/threads/thrzQIfbMYKI5yef9AoXWpIRg