hello,
I am using the example for the llamaspeak web from nvidia ai.lab,
the riva ASR works fine but I find it hard to talk to because it will recognise the end of speech very quickly even after short pauses in speech. is there a way to make the end of speech timeout longer? I didn’t find such a parameter while looking in the doc.
Hi @guy41 ,
I believe this should help you
https://docs.nvidia.com/deeplearning/riva/user-guide/docs/asr/asr-pipeline-configuration.html#beginning-end-of-utterance-detection
thanks! that looks like exactly what I was looking for
Hi again, sorry if it is a novice question since I am not well versed with riva
I am using riva_quickstart_arm64_v2.15.0 as shown in the example
which means that I used riva_init.sh that is inside and then riva_start.sh to start it.
I will also attach those files for reference.
Inside those I am unable to find where to change those parameters, it seems that there is no riva-build in those scripts. does that mean that this example uses an already built riva model and I need to build a new one to change the parameters?
riva_init.txt (9.0 KB)
riva_start.txt (4.7 KB)
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.