silero-vad
yarn
| silero-vad | yarn | |
|---|---|---|
| 10 | 4 | |
| 2,866 | 1,149 | |
| - | - | |
| 6.9 | 7.1 | |
| 11 days ago | 15 days ago | |
| Python | Python | |
| MIT License | MIT License |
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.
silero-vad
-
New models and developer products announced at OpenAI DevDay
>How do you detect speech starting and stopping?
https://github.com/snakers4/silero-vad
-
[Discussion] Video Translation Task
you could look into https://github.com/guillaumekln/faster-whisper especially the VAD section (Voice Activity Detector) using https://github.com/snakers4/silero-vad
-
Using Whisper to transcribe the entire Forensic Files series
I also had the same synchronization issue, so I wrote a WebUI/CLI that uses Silero-VAD that first splits the audio whenever there a silent portion (or every 30 seconds), and I haven't experienced it since:
-
Whisper - A new free AI model from OpenAI that can transcribe Japanese (and many other languages) at up to "human level" accuracy
By the way, I've updated the WebUI to now also support using Silero VAD to break up the audio into distinct sections, and run Whisper on each section and then combine them into one single transcript/SRT file.
-
[P] A more detailed post about Silero VAD on The Gradient
The VAD is always available on Github
- Silero VAD: pre-trained enterprise-grade voice activity detector
-
[P] Silero VAD: One voice detector to rule them all
I also pinned some interesting comments here regarding mobile and IOT usage here - https://github.com/snakers4/silero-vad/issues/37
- One voice detector to rule them all
yarn
-
Has Llama-3 just killed proprietary AI models?
There are techniques to extend the context window via fine tuning.
The authors claim this method was used to extend Llama 2 to 128k: https://github.com/jquesnelle/yarn
-
New models and developer products announced at OpenAI DevDay
Open researchers are trying to shrink and speed up 138K models e.g. YaRN https://github.com/jquesnelle/yarn
It's very compelling and opens up a lot of use cases, so I've been keeping an eye out for advancements.
-
Inf: Two 32k llama 70B models/adapters are available on huggingface
There have been existing extended context Lora's, I don't think yarn was implemented in heavily used frameworks yet, though.
- Yarn: Efficient Context Window Extension of Large Language Models
What are some alternatives?
whisper - Robust Speech Recognition via Large-Scale Weak Supervision
cheetah - On-device streaming speech-to-text engine powered by deep learning
kaldi-active-grammar - Python Kaldi speech recognition with grammars that can be set active/inactive dynamically at decode-time
GassistPi - Google Assistant for Single Board Computers
mr-robot - A multi-utility discord bot. Playback hilarious voice tracks on-demand, wiki for anything, turn on/off IoT enabled devices, and more!
hollow-knight-voice-commands - A fun little python tool to play Hollow Knight with only voice commands
mycroft-precise - A lightweight, simple-to-use, RNN wake word listener
Caster - Dragonfly-Based Voice Programming and Accessibility Toolkit
faster-whisper - Faster Whisper transcription with CTranslate2
spokestack-python - Spokestack is a library that allows a user to easily incorporate a voice interface into any Python application with a focus on embedded systems.
whisper-auto-transcribe - Auto transcribe tool based on whisper
autosrt - Offline srt producer gui with whisper.cpp