Whisper
git-bug
Our great sponsors
Whisper | git-bug | |
---|---|---|
32 | 56 | |
7,182 | 8,003 | |
- | - | |
6.5 | 6.3 | |
7 months ago | 5 days ago | |
C++ | Go | |
Mozilla Public License 2.0 | GNU General Public License v3.0 only |
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.
Whisper
-
Nvidia Speech and Translation AI Models Set Records for Speed and Accuracy
I've been using WhisperDesktop ( https://github.com/Const-me/Whisper ) with great success on a 3090 for fast & accurate transcription of often poor quality euro-english hours long multispeaker audio files. If there's an easy way to compare I'm certainly going to give this a try.
-
AMD's CDNA 3 Compute Architecture
Why would you want OpenCL? Pretty sure D3D11 compute shaders gonna be adequate for a Torch backend, and they even work on Linux with Wine: https://github.com/Const-me/Whisper/issues/42 Native Vulkan compute shaders would be even better.
Why would you want unified address space? At least in my experience, it’s often too slow to be useful. DMA transfers (CopyResource in D3D11, copy command queue in D3D12, transfer queue in VK) are implemented by dedicated hardware inside GPUs, and are way more efficient.
-
Amazon Bedrock Is Now Generally Available
https://github.com/ggerganov/whisper.cpp
https://github.com/Const-me/Whisper
I had fun with both of these. They will both do realtime transcription. Bit you will have to download the training data sets…
-
Why Nvidia Keeps Winning: The Rise of an AI Giant
Gamers don’t care about FP64 performance, and it seems nVidia is using that for market segmentation. The FP64 performance for RTX 4090 is 1.142 TFlops, for RTX 3090 Ti 0.524 TFlops. AMD doesn’t do that, FP64 performance is consistently better there, and have been this way for quite a few years. For example, the figure for 3090 Ti (a $2000 card from 2022) is similar to Radeon RX Vega 56, a $400 card from 2017 which can do 0.518 TFlops.
And another thing: nVidia forbids usage of GeForce cards in data centers, while AMD allows that. I don’t know how specifically they define datacenter, whether it’s enforceable, or whether it’s tested in courts of various jurisdictions. I just don’t want to find out answers to these questions at the legal expenses of my employer. I believe they would prefer to not cut corners like that.
I think nVidia only beats AMD due to the ecosystem: for GPGPU that’s CUDA (and especially the included first-party libraries like BLAS, FFT, DNN and others), also due to the support in popular libraries like TensorFlow. However, it’s not that hard to ignore the ecosystem, and instead write some compute shaders in HLSL. Here’s a non-trivial open-source project unrelated to CAE, where I managed to do just that with decent results: https://github.com/Const-me/Whisper That software even works on Linux, probably due to Valve’s work on DXVK 2.0 (a compatibility layer which implements D3D11 on top of Vulkan).
-
Ask HN: What is your recommended speech to text/audio transcription tool?
Currently, I use a GUI for Whisper AI (https://github.com/Const-me/Whisper) to upload MP3s of interviews to get text transcripts. However, I'm hoping to find another tool that would recognize and split out the text per speaker.
Does such a thing exist?
- Da audio a testo, consigli?
-
Ask HN: Any recommendations for cheap, high-quality transcription software
I just used Whisper over the weekend to transcribe 5 hours of meeting, worked nicely and it can be run on a single GPU locally. https://github.com/ggerganov/whisper.cpp
There are a few wrappers available with GUI like https://github.com/Const-me/Whisper
- Voice recognition software for German
- Const-me/Whisper: High-performance GPGPU inference of OpenAI's Whisper automatic speech recognition (ASR) model
- I built a massive search engine to find video clips by spoken text
git-bug
-
Radicle: Peer-to-Peer Collaboration with Git
Unfortunately github appears to be actively breaking the ability to use git-bug on large repositories (like nixpkgs):
https://github.com/MichaelMure/git-bug/issues/749#issuecomme...
-
Nintendo emulator 'Suyu' removed from Gitlab following DMCA request
True but getting less true by the day:
https://github.com/MichaelMure/git-bug
https://www.fossil-scm.org/home/doc/trunk/www/index.wiki
- CRDTs Turned Inside Out
-
Sourcehut and Codeberg are both currently experiencing a DDoS attack
Only not having access to https://todo.sr.ht made me to recognize fully, that I don’t have any access to it. https://github.com/MichaelMure/git-bug suddenly looks much more interesting.
-
Gothub: Alternative front-end for GitHub written with Go
Neither do the issues support. But there is git-bug [0].
[0]: https://github.com/MichaelMure/git-bug
-
git-appraise – Distributed Code Review for Git
As a sort of spiritual successor to git-appraise, I've been working on git-bug[1] which support issues and will at some point support kanban and code review. There is a few notables improvements:
- CRDT-like reusable data structure [2][3] for true p2p workflow and easily create new entities (code review ...)
- bidirectional bridges to github, gitlab ... to ease the transition or just use git-bug as a complement of those platform
- CLI, terminal UI and web UI, for different taste and integrate into your tooling/workflow
[1]: https://github.com/MichaelMure/git-bug
[2]: https://github.com/MichaelMure/git-bug/blob/master/doc/model...
[3]: https://github.com/MichaelMure/git-bug/blob/master/entity/da...
-
Show HN: Gitopia: Decentralized GitHub Alternative for Open Source Collaboration
> but that is for the development of the platform and network of Gitopia. For the end user the workflows remain almost the same for collaboration.
I have to disagree here. Accidental complexity in a system can have severe downstream impacts on end users, whether that be in the form of poor performance, unreliability, or just slow update cycles. It's not something you can paper over and completely hide from the user.
> Along with this the blockchain layer layer offers immutable, transparent and tamper proof versioning of code
Tamper-proof can be accomplished natively by signing [0]. receive.denyNonFastForwards and receive.denyDeletes[1] can be used to make a git repository immutable. Git commits are also already content-addressable. And transparency is achieved by just having the repo available for people to clone.
> along with the collaboration meta and augments the current collaboration flow
Could this augmentation not be accomplished by storing the collaboration information in the repo under a set of special-purpose branches? Like git-bug[2] or git-issue[3]? Coupled with GPG signatures and you've got your immutability, too!
> Along with this it enables us to provide a novel means to incentivize open-source contributions along with fostering a more decentralized approach for governance (even for projects), every token holder could have a say in the decision making, reducing the risk of undue influence by a single party, hence eliminating centralized control.
This one I'll grant you, but it's by far the least compelling aspect of the project to me. I don't think we're going to solve the centralization of GitHub by centralizing on a new plutocracy, I'd much rather see efforts towards full decentralization. There's nothing inherent to Git that requires that we all use the same set of servers.
[0] https://git-scm.com/book/en/v2/Git-Tools-Signing-Your-Work
[1] https://git-scm.com/book/en/v2/Customizing-Git-Git-Configura...
[2] https://github.com/MichaelMure/git-bug
[3] https://github.com/dspinellis/git-issue
-
So, I went down the rabbit hole of buying GitHub Stars, so you won't have to
Regarding the issues, there are some projects like git-bug https://github.com/MichaelMure/git-bug trying to embed these sorts of meta-work into git.
-
Let's Make Sure Github Doesn't Become the only Option
Probably git-bug is closer to what Fossil does: It uses Git as a storage engine, and can coexist with your code in the same physical repository, but the issues don't actually show up as source files. Instead, each issue is a special branch (buried in refs so it won't clutter up git branch) that has zero common ancestry with anything else. So in theory you can poke at it with Git, but really, the Git under the hood is mostly an implementation detail, and as long as you interact with those files through the tool, it guarantees you won't have merge conflicts.
-
Clocks and Causality – Ordering Events in Distributed Systems
You might be interested by git-bug and https://github.com/MichaelMure/git-bug/blob/master/doc/model..., which seems to be exactly what you describe. (Disclaimer: author).
What are some alternatives?
whisper.cpp - Port of OpenAI's Whisper model in C/C++
git-issue - Git-based decentralized issue management
whisper - Robust Speech Recognition via Large-Scale Weak Supervision
EdenSCM - A Scalable, User-Friendly Source Control System. [Moved to: https://github.com/facebook/sapling]
TransformerEngine - A library for accelerating Transformer models on NVIDIA GPUs, including using 8-bit floating point (FP8) precision on Hopper and Ada GPUs, to provide better performance with lower memory utilization in both training and inference.
nessie - Nessie: Transactional Catalog for Data Lakes with Git-like semantics
just-an-email - App to share files & texts between your devices without installing anything
Kaiserreich-4-Bug-Reports - Issue tracker for Kaiserreich for Hearts of Iron 4
ggml - Tensor library for machine learning
dolt - Dolt – Git for Data
beaker - An experimental peer-to-peer Web browser
gumtree - An awesome code differencing tool