doing something
@doing_something
710 subscribers
909 photos
36 videos
36 files
1.94K links
@smlkw doing something, just my notes to keep updates
Download Telegram
About
Blog
Apps
Platform
Join
doing something
710 subscribers
doing something
Ukrainian voices for Text-to-Speech now in separated datasets:

https://huggingface.co/collections/speech-uk/ukrainian-text-to-speech-67bd059d61b2598f3a2a7969

#ai #speech
huggingface.co
Ukrainian Text-to-Speech datasets - a speech-uk Collection
Five voices: Mykyta, Oleksa, Lada, Kateryna or Tetiana
🔥2
583 views
doing something
Added a section with phonemizers to https://github.com/egorsmkv/speech-recognition-uk

https://github.com/egorsmkv/speech-recognition-uk/issues/57

#ai #speech
GitHub
GitHub - egorsmkv/speech-recognition-uk: 🇺🇦 Speech Recognition & Synthesis for Ukrainian
🇺🇦 Speech Recognition & Synthesis for Ukrainian. Contribute to egorsmkv/speech-recognition-uk development by creating an account on GitHub.
317 views
doing something
https://github.com/huggingface/setfit

#ai #nlp
GitHub
GitHub - huggingface/setfit: Efficient few-shot learning with Sentence Transformers
Efficient few-shot learning with Sentence Transformers - huggingface/setfit
124 views
doing something
Let's cook, guys

#ai #speech
🔥1
124 views
doing something
https://burn.dev/blog/why-quantization-matters/

#ai #optimization
burn.dev
Why Quantization Matters
Modern deep learning models, such as large language models (LLMs), are heavily constrained by memory bandwidth. GPUs can execute floating-point operations (FLOPs) much faster than they can fetch weights from memory. For instance, an NVIDIA A10 has a peak…
134 views
doing something
https://huggingface.co/blog/Steveeeeeeen/llasagna

#ai #speech #tts #vc
huggingface.co
From Llasa to Llasagna 🍕: Finetuning LLaSA to generates Italian speech and other languages
A Blog post by Steven Zheng on Hugging Face
136 viewsedited  
doing something
doing something
Let's cook, guys #ai #speech
As always, it's #daily_pain
122 views
doing something
doing something
Let's cook, guys #ai #speech
It's done.

https://huggingface.co/datasets/speech-uk/voice-of-america

#ai #asr #speech
🔥2
411 views
doing something
https://t.me/neuroComputational
Telegram
Neural Encoder 🧠🌊📟
Spiky notes on computational neuroscience and biology 🔬🧬 ML / AI / AGI.
Let’s compute and fire together — so we can wire up cross-disciplinary topics.
124 views
doing something
Polars even helps to code

#ai #polars
❤2
119 views
doing something
Made a space to simplify my ASR research:

https://huggingface.co/spaces/Yehor/evaluate-asr-outputs

#ai #speech
👍1🔥1
436 views
doing something
https://huggingface.co/spaces/fastrtc/object-detection

https://huggingface.co/spaces/fastrtc/object-detection/blob/main/app.py

#cv #ai
huggingface.co
Object Detection - a Hugging Face Space by fastrtc
This app detects objects in real-time video streams. Users provide a live video feed, and the app highlights detected objects with confidence scores.
128 views
doing something
https://github.com/papers-we-love/papers-we-love
GitHub
GitHub - papers-we-love/papers-we-love: Papers from the computer science community to read and discuss.
Papers from the computer science community to read and discuss. - papers-we-love/papers-we-love
129 views
doing something
Batch inference using vllm

#ai #vllm #speech
124 views
doing something
https://arxiv.org/abs/2412.05265

#rl
arXiv.org
Reinforcement Learning: An Overview
This manuscript gives a big-picture, up-to-date overview of the field of (deep) reinforcement learning and sequential decision making, covering value-based methods, policy-based methods,...
129 views
doing something
doing something
Batch inference using vllm #ai #vllm #speech
Added batch mode calculation to https://huggingface.co/spaces/Yehor/evaluate-asr-outputs

Btw, batched vllm inference gives better RTF for Whisper

#asr #ai #speech
135 viewsedited  
doing something
GitHub Copilot already has new Sonnet models

#ai
138 views
doing something
vllm + A100 (40 GB in Google Colab) + https://huggingface.co/Yehor/whisper-large-v3-turbo-quantized-uk using batch_size=128

#ai #speech #whisper
🔥2
1.24K views
doing something
Initial work to add OWSM models from Espnet:

https://github.com/egorsmkv/speech-recognition-uk/issues/69

#ai #speech
GitHub
Add OWSM-CTC models · Issue #69 · egorsmkv/speech-recognition-uk
https://huggingface.co/collections/espnet/owsm-ctc-ultra-fast-speech-foundation-models-67ab7a9d91a82bab571ca42c
138 views
doing something
Forwarded from Yehor Smoliakov
Що ж, підготовка до інференсу великих датасетів готова.

Далі сам інференс та вирівнювання даних для створення якісного speech to text датасету.

Якщо бажаєте зробити донат на хмарні GPU, ось банка:

🔗Посилання на банку
https://send.monobank.ua/jar/3Saxixsdua

💳Номер картки банки
5375 4112 0249 3776

—

Результати будуть опубліковані під відкритими ліцензіями 😎
send.monobank.ua
Безпечний переказ коштів
Надсилайте безкоштовно та безпечно кошти
👍2
142 views