tevr-asr-tool
pyctcdecode
tevr-asr-tool | pyctcdecode | |
---|---|---|
9 | 2 | |
408 | 405 | |
0.0% | 0.7% | |
5.9 | 2.1 | |
over 1 year ago | 10 months ago | |
C | Python | |
MIT License | Apache License 2.0 |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
tevr-asr-tool
-
Ask HN: Why is pay so much higher in the US? (or is it?)
Do some cool open source stuff. Since August 9th, I received about 50 job offers mentioning this repo: https://github.com/DeutscheKI/tevr-asr-tool Most of them were senior engineer or AI researcher, with a few CTO / co-founder offers sprinkled in. I'm not in the market and this was a bit unexpected to me, but those emails sounded like they would pay well. And most was remote for US companies.
-
Ask HN: Who is using C++ as the main language for new project?
Me, because I plan to extend to realtime processing: https://github.com/DeutscheKI/tevr-asr-tool
Also, all the performance critical stuff on my production servers is C++ with JNI or pybind wrappers.
- Show HN: 用284行C++语言实现最先进的德语语音识别 (Show HN: State-of-the-art German speech recognition in 284 lines of C++)
-
Show HN: State-of-the-Art German Speech Recognition in 284 lines of C++
The unique work that makes this speech recognition superior to other tools is in those 284 lines of code: https://github.com/DeutscheKI/tevr-asr-tool/blob/master/tevr...
That's a custom-designed beam search decoder implemented in C++ and based on the research for my TEVR paper. It increases performance by a relative 16% reduction in word error rate.
pyctcdecode
-
Show HN: State-of-the-Art German Speech Recognition in 284 lines of C++
I wrote "284 lines of C++" to indicate that this is compact enough for people to actually read and understand the source code. Also, compiling my implementation is super easy and straightforward ... something which can't be said for Kaldi, Vosk, or DeepSpeech.
If you try to read the CTC beam search decoder from Mozilla's DeepSpeech [1], that alone is about 2000 LOC in multiple files.
If you try to read the pyctcdecode source that is used by HuggingFace [2], that's 1000+ LOC of Python.
But this implementation is all the client-side, i.e. the entire "native_client" folder hierarchy in DeepSpeech [3], narrowed down to a mere 284 lines.
[1] https://github.com/mozilla/DeepSpeech/tree/master/native_cli...
[2] https://github.com/kensho-technologies/pyctcdecode
[3] https://github.com/mozilla/DeepSpeech/tree/master/native_cli...
- kensho-technologies/pyctcdecode
What are some alternatives?
DeepSpeech - DeepSpeech is an open source embedded (offline, on-device) speech-to-text engine which can run in real time on devices ranging from a Raspberry Pi 4 to high power GPU servers.
k8deployer - An experimental deployer for kubernetes apps for developers who are too lazy (or busy) to learn Helm.
SLID-on-Microcontrollers - Speech Classification using a Convolutional Neural Network running on a Microcontroller
Typesense - Open Source alternative to Algolia + Pinecone and an Easier-to-Use alternative to ElasticSearch ⚡ 🔍 ✨ Fast, typo tolerant, in-memory fuzzy Search Engine for building delightful search experiences
MathAnimation - A simple C++/OpenGL application to create quick and dirty mathematically accurate animations
materialize - The data warehouse for operational workloads.
gui_starter_template - A template CMake project to get you started with C++ and tooling
CppCoreGuidelines - The C++ Core Guidelines are a set of tried-and-true guidelines, rules, and best practices about coding in C++
awesome-modern-cpp - A collection of resources on modern C++
windmap
wPedals - Guitar Pedals Widgets For The Web
litestream - Streaming replication for SQLite.