-
Hi,
In icefall, there are multiple decoding methods available, eg. greedy_search, beam_search, modified_beam_search, fast_beam_search, fast_beam_search_nbest. There are some other decoding methods …
-
PyTorch 2.0 is now available which now includes two major things that may be useful for faster training:
1. TorchDynamo, which can be used by calling `torch.compile(model)`
2. FLASH attention, by us…
-
Hi,
I found your bk5000.py setup and it works very well in collecting a real-time stream from the BK5000.
However, I am new to bytes and TCP links and I try to figure out how to decode the recei…
-
With the k2 models, is it possible to output phone-alignment and word-alignment info along with confidence like the old kaldi models?
-
A lot of the bluebird feature requests deal with lists and iterables and various methods to execute them. Proposals include sequential iteration, concurrency controls (#1100), infinite list iteration …
spion updated
6 years ago
-
How is this step applied in detail?
Build sherpa-ncnn (C++ code)
Link
https://k2-fsa.github.io/sherpa/ncnn/android/build-sherpa-ncnn.html
-
使用命令行生成中文字热词文件后,在tokens.txt能找到对应的byte,但是使用时有异常
![微信图片_20240508112931](https://github.com/k2-fsa/sherpa-onnx/assets/166801697/a391aab4-3442-48ee-819e-8e80c56e3e39)
![微信图片_20240508113007](https://gith…
-
When I see some of the topics here and elsewhere about building languages for rust, most of it become talks about performance, deals with quirks of platforms, GC, and other matters that could obscure…
mamcx updated
4 years ago
-
Hi Josh,
I've been moving toward MCMC results being in the form of an iterator instead of an array, and encouraging others in this direction as well. This convenience and flexibility in a lot of di…
-
Create a issue here to discuss this problem.
I re-ran the pruned_transducer_stateless4 baseline with `DynamicBucketingSampler`, the results are as follows:
setup | greedy_search | fast_beam_sear…