Shayne Mei

Support for VAD / endpointing

7

Does the sherpa server support VAD or endpointing?

Return alignment info

5

With the k2 models, is it possible to output phone-alignment and word-alignment info along with confidence like the old kaldi models?

How does the `nn_pool_size` affect concurrency? What are we expected to see when we tune this parameter? I ran several concurrency latency tests with different `nn_pool_size` (number of workers) on...

Performance gap between icefall local streaming decoding and sherpa streaming decoding

12

Using the same model (a streaming pruned_transducer_stateless5 trained on gigaspeech), we are experiencing some performance gap between local icefall streaming decoding and sherpa server streaming decoding. WERs for both setup...

Latency for pruned transducer stateless5 streaming models

30

We are trying out the pruned transducer stateless5 streaming models and we're experiencing unstable/increasing latency in streaming when sending 0.1s chunks of audio. What is the expected latency for these...

Is it possible to obtain past trending searches given a datetime?

Use explicit relative imports

2

- switch from implicit relative imports to explicit relative imports as recommended by Guido: https://peps.python.org/pep-0328/#guido-s-decision - sort import order using `isort` close #508

pre-trained gigaspeech model decoding error

13

Unable to decode AMI SDM full-corpur-asr dev set. 1. Prepared AMI test and dev set using this command: ```bash lhotse prepare ami \ --mic sdm \ --partition full-corpus-asr \ \...

How would you recommend finetuning the model with a language model to help recognize certain words using the context? In other words, is there a way to increase the weight...

Streaming server printing empty lines

5

When using the endpointer, the streaming server would print empty lines like this:  There seems to be one empty line per segment (whenever...