TTS Streaming Delayed Until Conversation Agent Completes #2877
Unanswered
mikenorgate
asked this question in
Voice assistants
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
Describe the feature
When using streaming conversation agents with TTS, the audio playback is delayed until the entire conversation response completes, even though text is available and streaming to the TTS engine immediately.
When creating complex assist pipelines there can be a fairly large delay between it starting and finishing, being able to give status updates as the flow is progressing would be really useful
Example commands
N/A
Use cases
I have used a webhook conversation integration to connect assist to n8n, this will run some flows to either take direct actions or redirect to an LLM for more conversational queries.
Due to some of the flows being slow, like waiting for external services, it want to be able to send updates while it's processing. e.g.
One moment while I process your request...Anything else?
I have got this working up to a point. I have the conversation agent sending back multiple responses correctly however TTS doesn't kick in until the last message is received and the webhook connection is closed.
From what I can see from the code around this the processing currently synchronous, the conversation agent has to complete before TTS is run, even though with streaming enabled it sends the text to the TTS buffer as it's received .
My current series of events looks like this, based on my rough debugging of the pipeline
Beta Was this translation helpful? Give feedback.
All reactions