Hi @nithinraok ,
Seeking guidance in inferencing using ASR_Contenxt_Biasing for the above streaming.
The tutorial shows only evaluation script. Also , the cache based streaming asr doenst have punctuations . Is there any argument to be activated to enable punctuations ?

Current published cache-aware models have no punctuations, we are working on an updated version.
Maybe do streaming and then update punctuations in the post-processing stage once model detects . ?
CC: @andrusenkoau for context biasing using streaming settings/models.

Hi @AbhijithMallya,
The current solution for word boosting works only in offline scenarios. It does not support streaming for now.

Hi @andrusenkoau ,
Is it possible to add VAD on top of cache_based_streaming ?
CC: @nithinraok

Hi @AbhijithMallya, The current solution for word boosting works only in offline scenarios. It does not support streaming for now.

Hi @andrusenkoau,
I wanted to know if It is possible to fine tune the stt_streaming models to a indian languages , using the training scripts and tutorials provided in the nemo documentation ? .

CC : @nithinraok

Yes, it is absolutely possible to implement a WebSocket-based interface for exposing a streaming ASR (Automatic Speech Recognition) endpoint using the Cache Aware Streaming approach shown in Online_ASR_Microphone_Demo_Cache_Aware_Streaming.ipynb.

if you find this is right then leave me a like and mark me the answer thankyou!