Ha! Semantics – in this case we literally force it to transcribe into a target language. So, technically you can say it is translating! Whisper traditionally only translates from language “X” into English.
@reach_vb
-
AI Video Generation from Audio Using Word-Level Timestamps
By
–
The audio snippet was taken from "Stupid People" comedy routine by Bill Engvall The video was generated via @mhollemans
's brilliant Word-level timestamps space: -
Testing Model Performance on Untrained Tasks: Benchmarking Results
By
–
Quick caveat: It's a task that the model isn't explicitly trained on, so test it thoroughly before us We'll run more benchmarks this week and share our findings soon! Help us test it better? 🙂
-
Whisper Hack Enables Multilingual Transcription Beyond English
By
–
We found a hack to make Whisper transcribe into *any* language (not just English)! ⚡️
— Vaibhav (VB) Srivastav (@reach_vb) 26 juin 2023
This is an EXPERIMENTAL one-step feature: enter your audio as input, and get a transcription in another language. 🔥
Check it out here: https://t.co/afgeUfw2xF pic.twitter.com/7DoJ93Z2ZBWe found a hack to make Whisper transcribe into *any* language (not just English)! This is an EXPERIMENTAL one-step feature: enter your audio as input, and get a transcription in another language. Check it out here: https://
github.com/Vaibhavs10/tra
nslate-with-whisper
… -
Converting ArXiv Citations to Accurate Publication References
By
–
Does anyone have any recommendations for converting Arxiv citations to the most accurate publication citations? I have ~ 350 references, some from DBLP, some ACL but most are arxiv citations, looking for a way to find the most up-to-date citation for them? Help pls!?
-
MMS Integration in Transformers: Easy Inference and Fine-Tuning
By
–
5 / 5 We've integrated MMS into Transformers making the models very easy to use for both inference and fine-tuning. Demo https://
huggingface.co/spaces/faceboo
k/MMS
…
Docs https://
huggingface.co/docs/transform
ers/main/en/model_doc/mms
…
Fine-Tuning -
MMS AI Transcribes 1000+ Languages Including Endangered Tongues
By
–
3 / 5 MMS is capable of transcribing 1,000+ languages, many of which are endangered, such as Ari or Kaivi. In the future, MMS can play a vital role in keeping languages alive by helping the remaining speakers to create written records and communicating in their native tongue.
-
MMS Adapters: Training Efficient Multilingual Language Models
By
–
4 / 5 To adapt to 1000+ different vocabularies, MMS uses Adapters – a training method where only a small fraction of model weights are trained Adapter layers act like linguistic bridges, enabling the model to leverage knowledge from one language when deciphering another.
-
3000 Living Languages Endangered by Globalization Trends
By
–
2 / 5 According to https://
ethnologue.com around 3000, or 40% of all "living" languages, are endangered due to fewer and fewer native speakers.
This trend will only continue in an increasingly globalized world. -
Meta’s Multilingual Speech Model Democratizes AI Access Globally
By
–
Meta AI's recently released "Massively Multilingual Speech" (MMS) model is a huge step forward towards democratizing
Speech to every corner of the globe. In addition, it might also play a significant role in preserving global linguistic diversity. https://
huggingface.co/papers/2305.13
516
…