Hi, I'm encountering an issue where I'm getting the first 3 seconds of audio from speech to text getting cut off. I've tried running both a remote (running on a high end desktop) and local piper instance (thinking it might be a performance issue from running piper on the RP4) and it happens with both. Has anyone encountered this issue and knows a workaround to delay the audio tts playback so that maybe it doesn't cut off?
#Calling tts say with Piper TTS cuts off the first 3 seconds
thin nest
woeful frigate
no. works fine with esp32 as target device.
i'm using this
action: tts.speak
data:
media_player_entity_id: media_player.esp32_s3_media_player
message: This is a test message
cache: false
target:
entity_id: tts.piper
But if i'm using my nvidia shield as the target. the beginnig is missing 😉
thin nest
Ah I see. I have a couple of S3's I haven't setup, but my setup revolves around HASS.agent's running on several media pcs, each connected through a surround sound receiver to broadcast the audio through in ceiling speakers.
On several of my media pcs running on nucboxes I'm getting consistent audio cut offs in the beginning of the tts
No delay on my main pc. I also tried to increase the thread priority on the HASS.agent process to realtime thinking it was a performance issue with that, but no luck either
When I use mimic3 + maryTTS, there is no delay, only when using piper specifically (and I tried two different instances of piper)
The problem with the latter setup, is that I can't seem to get it configured to work with the assist pipeline. So no matter which way I configure it, something isn't working right :/
Also to clarify, both machines are on local wired ethernet, in case this was an issue with internet delay
Also, thank you for the response!
Also tried running the latest piper image with the same result:
