Closed sdetweil closed 8 months ago
I'm confused. The dashes have always been there. The very first commit has them: https://github.com/rhasspy/wyoming/blob/892e01dea82ccb804d93a25dc8eef74a2d4fca37/wyoming/audio.py#L11
How are you using Wyoming?
weird indeed. I was looking at what I thought was 1.5 (master), but using 1.3 and couldnt get the is_type to work. at that point I wasn't using the class libs. so this was probably my fault. transcribe however was different. so I ass-umed.
sorry, didn't answer the second question..making a new asr. mostly cause I want to use it with an existing app but needed a way to test it without all the HA baggage.. so I needed a client that can send a wav file that the asr can transcribe
the doc says 'asr that anything can connect to' .. so as I don't know python nearly at all, I chose nodejs , no response .. bash nc next (but that had the connect/disconnect problem) , but I could send the tranactions then python telnetlib, which stayed connected and did all the transactions except AudioChunk, samples showed needing a newline.. the nodejs trouble was the lack of a newline..
then gave up and cobbled together a python wyoming client.. and I don't know much about audio streams, or files.. so its all fun times..
Yeah, I had misspelled "transcribe" early on :man_facepalming:
Anything that can do TCP and JSON should be able to use Wyoming. It is important that the TCP connection be maintained through the session. I don't know much about nodejs, but I'm surprised it had problems :/
nodejs didn't have trouble, just needed to explicitly send a newline. then the server side responded. no info before that
closing, not an issue
change from audiostop to audio-stop
what useful value is that?