RIVA node JS GRPC

AI & Data Science Deep Learning (Training & Inference) Riva
1

Please provide the following information when requesting support.

Hardware - GPU (A100/A30/T4/V100) Quadro RTX 8000
Hardware - CPU Intel(R) Xeon(R) Gold 5218R CPU @ 2.10GHz
Operating System - Ubuntu 18.04.5 LTS
Riva Version - 1.9 beta
TLT Version (if relevant)
How to reproduce the issue ? (This is for errors. Please share the command and the detailed log here)
Hi, I was able to start communicating with the RIVA server via node JS GRPC, however after sending the streaming config, I am getting a “Releasing idle sequence with correlation” in the docker logs
Does anyone know what that means ?
Attached below is an excerpt of ‘docker logs riva-speech’ :
I0329 08:11:45.462642 11748 grpc_riva_asr.cc:913] ASRService.StreamingRecognize called.
I0329 08:11:45.462730 11748 grpc_riva_asr.cc:947] ASRService.StreamingRecognize performing streaming recognition with sequence id: 2103500628
I0329 08:11:45.462811 11748 grpc_riva_asr.cc:1001] Using model riva-asr for inference
I0329 08:11:45.462989 11748 grpc_riva_asr.cc:1017] Model sample rate= 16000 for inference
I0329 08:11:45.463202 11748 riva_asr_stream.cc:213] Detected format: encoding = 1 RAW numchannels = 1 samplerate = 16000 bitspersample = 16
W:batch_slot_manager.cc:105: Releasing idle sequence with correlation id 1329347422 idle time 557587642us
W:batch_slot_manager.cc:105: Releasing idle sequence with correlation id 1968383396 idle time 510828680us

1 Like
2

For more context, this is the NodeJS client code I am using based on a post made earlier :
Riva : Node JS examples
I also attached my config.sh file for reference :
config.sh (7.5 KB)

3

Sounds like you might just need to close the gRPC client when done?
There’s a close() method for the RivaSpeechRecognitionClient that you create.

4

Hey @pineapple9011, just fyi, I also had a nodejs grpc client set up and running when you made that NodeJS example post. I did not use protoloader as you did and used another library for generating ts bindings for the client. However, unlike you I never ran into the issue of having to trim out the extra header at the start of the wav file.

5

After closing the stream and client, I now receive a ASRService.StreamingRecognize returning OK in my docker logs.
Does that mean that my transcription is successful and if so how do I get the results?
In addition, is there a need to change the sample rate of my wav files to the one set in the streaming config, and is there a max duration supported for the wav files (e.g 3 minutes and above)
Attached below is my docker logs riva-speech :
I0331 07:49:24.104351 5752 grpc_riva_asr.cc:913] ASRService.StreamingRecognize called.
I0331 07:49:24.104424 5752 grpc_riva_asr.cc:947] ASRService.StreamingRecognize performing streaming recognition with sequence id: 1590625615
I0331 07:49:24.104507 5752 grpc_riva_asr.cc:1001] Using model riva-asr for inference
I0331 07:49:24.104668 5752 grpc_riva_asr.cc:1017] Model sample rate= 16000 for inference
I0331 07:49:24.104849 5752 riva_asr_stream.cc:213] Detected format: encoding = 1 RAW numchannels = 1 samplerate = 16000 bitspersample = 16
I0331 07:49:24.944602 6046 grpc_riva_asr.cc:674] Send silence buffer for EOS
I0331 07:49:32.039855 5752 grpc_riva_asr.cc:1105] ASRService.StreamingRecognize returning OK

6

I never ran into the issue of having to trim out the extra header at the start of the wav file.

Interesting, what version of Riva was this?
I don’t think we had to do this in newer versions, but I never checked.

7

Does that mean that my transcription is successful and if so how do I get the results?

Yup, the logs you have are what I get as well. Results should be in your response callback. If you’re using streams, it’s the data callback.

Is there a need to change the sample rate of my wav files to the one set in the streaming config

Of course!

Is there a max duration supported for the wav files

Yeah, I think you specify that when you build your model, check the docs. Not sure what the default examples use though.

8

Thanks for the help! Managed to get it working.

9

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.