TTS response doesn't play first time

I’m using a HA Voice PE device but I also had this same issue with an Atom Echo.

If the device has been sitting idle for a while, the first use does not give an audible response.

For example, I say “ok nabu”… it lights up and dings… I say “what time is it”… the light goes off, and then no response. Do the same thing again right after and it answers “4:30 pm”.

Same situation, if I ask it to do something (like add an item to a list), the item does get added, just no audible response. But then do it right after, and the item gets added, and I do get the audible response.

This is pretty consistent behavior, though couldn’t say for sure it’s like this every time (it may be that occasionally I will hear the answer the first time).

I’m using Home Assistant Cloud for TTS and Speech-to-text, and when I open the debug panel for it, both the no-response and heard-response entries look identical and successful. Specifically TTS shows successful, and I can play the TTS response from the debug panel for both entries.

HA is running on an HA Green device.

Anyone have any idea what is happening? Is there any other place I can look for information to debug this?

Thanks in advance for an help here.

I get some latency on the first command too - though only on commands that require a reply. Simple “turn on/off” commands are immediate. I always get the reply after a short pause, though.

Since it only happens with spoken replies, I had assumed that it was something to do with TTS caching.

It seems to be less apparent with some TTS services - quite pronounced with Amazon Polly, hardly happens at all with Elevenlabs. You might try experimenting.

I figured out what the issue was.

I was using duckdns for the connection, and it was unreliable. Now using default non-https for local connections and it’s working reliably.