robocrunch
Patrick von Platen
@PatrickPlaten
Speech 🗣️ @Huggingface
Tweets by Patrick von Platen
https://apply.workable.com/huggingface/j/146F461743/
Speech News📢 We just opened two new exciting positions for deep learning in speech🔥 👉 Research Engineer to pretrain large audio Models: htt ...
Shared by
Patrick von Platen
at
5/3/2022
2-0
https://huggingface.co/models?other=data2vec-audio
Audio model beats Wav2Vec2 on Librispeech: https://t.co/h2F2cHRz52 Vision model beats Beit on ImageNet: https://t.co/MOr4Y6qyAG
Shared by
Patrick von Platen
at
4/21/2022
0-3
https://huggingface.co/docs/transformers/main/en/model_doc/data2vec#overview
🧙♂️One pre-training objective to rule all modalities... We've recently merged @MetaAI's new Data2Vec model into🤗Transformers. Check it out ...
Shared by
Patrick von Platen
at
4/21/2022
2-17
https://arxiv.org/abs/2203.10752
Very excited to release the first FULLY accessible multi-lingual speech benchmark! https://t.co/jBO8PHIBnK
Shared by
Patrick von Platen
at
3/31/2022
0-7
https://huggingface.co/spaces/huggingface/hf-speech-bench
ALL speech recognition models in ALL languages in ONE🤗Space. Check out the new best place to find SOTA speech recognition models in the langua ...
Shared by
Patrick von Platen
at
3/25/2022
12-36
Text-generation has a guide!🦮 @cwkeam did an incredible job at adding constrained beam search to🤗Transformers. It allows you to force text-g ...
Shared by
Patrick von Platen
at
3/15/2022
1-6
https://huggingface.co/models?other=voxpopuli-v2
🎙️Speech news 📢 We've added 31 new pretrained mono-lingual Wav2Vec2 checkpoints from @MetaAI and @ChanghanWang to the Hub🤗 Checkpoints are ...
Shared by
Patrick von Platen
at
2/28/2022
6-22
https://huggingface.co/google/t5-efficient-tiny-nl8
Every checkpoint is available in PyTorch, Tensorflow and JAX and comes with an in-detail model card. The new checkpoints are especially intere ...
Shared by
Patrick von Platen
at
2/16/2022
1-4
https://github.com/kensho-technologies/pyctcdecode
🗣️ Speech news: LM boosted ASR We're finally starting to have better support for LM-boosted speech recognition🔥 🤗Transformers' Wav2Vec2 + @ ...
Shared by
Patrick von Platen
at
11/4/2021
31-139
https://huggingface.co/models?other=t5-lm-adapt
T5's LM adapted checkpoints have just been uploaded to the 🤗 Hub! The checkpoints work very well for prompt tuning and were used for @Bigscien ...
Shared by
Patrick von Platen
at
11/1/2021
24-146
https://arxiv.org/pdf/2110.01900.pdf
HuBERT by @facebookai has a new student🧑🎓 DistilHuBERT by @ntu_spml is the most efficient Speech model yet. The model keeps most of HuBERT ...
Shared by
Patrick von Platen
at
10/28/2021
39-195
https://angel.co/l/2vTiUi
We are looking for a research intern in speech recognition @huggingface🎙️ If you're excited about pre-trained Transformer models for speech re ...
Shared by
Patrick von Platen
at
10/1/2021
39-164
https://twitter.com/anton_lozhkov/status/1435256778657304580
Wav2Vec2 and Hubert can now be used for audio-classification in 🤗Transformers!
Shared by
Patrick von Platen
at
9/7/2021
1-7
https://arxiv.org/abs/2104.01027
ASR models trained on read-out audio alone are often not practical for real world usage. Robust Wav2Vec2 (https://t.co/T0KZJNq7VF) by @facebook ...
Shared by
Patrick von Platen
at
8/28/2021
58-238
https://colab.research.google.com/drive/1qn7d9FkEOEIQcaLr2WFhopH64JGKyXe6?usp=sharing
The engine of language model prompting is a fast auto-regressive generation function! We've recently added auto-regressive generation in Flax ...
Shared by
Patrick von Platen
at
6/18/2021
33-175
https://twitter.com/huggingface/status/1405202422813843464
Welcome HuBERT to the transformers Speech family!
Shared by
Patrick von Platen
at
6/16/2021
7-60
