Nikko Strom retweeted
🗣️ Announcing Amazon Nova Sonic, a new speech-to-speech foundation model that can understand voice as input & generate a human-like voice as output.
➡️ Available via a new API in Amazon Bedrock, the model simplifies the development of voice applications: amzn.to/4cljCZ7
Looking for an off-ramp. Is there a mastodon server with an active ML/AI discussion? @NandoDF @Miles_Brundage @ilyasut @Smerity @ctnzr @karpathy @goodfellow_ian
2
You know that neural net weights tend to be Gaussian distributed, right? And you probably have an idea why, right? But do they have to be? In this paper we show that, no they don't, and this can be very useful when quantizing weights aggressively. amazon.science/publications/…
1
Nikko Strom retweeted
Today, Amazon took three steps toward letting people interact with voice AI systems in their native tongues, announcing (1) a 51-language parallel dataset (MASSIVE), (2) a new competition using that dataset (MMNLU), and (3) a related workshop at EMNLP '22. amzn.to/3MlBOn9
Deep learning is a ball of mud. #AmazonScience
What’s next for deep learning? amazon.science/blog/whats-ne…
Alexa is taking another leap towards natural conversation. This is rolling out gradually on Show 10 now, so please have patience if your device is not enabled yet. #Alexa #AlexaAI #ComputerVision #AmazonScience amazon.science/blog/new-alex…
1
Nikko Strom retweeted
On the next episode of Alexa & Friends, @jeffblankenburg will be chatting with Pradeep Natarajan, Alexa AI principal applied scientist, about his work in computer vision, machine learning, and multimedia processing. Tune in at 9am PT on Oct. 28. #Alexa amzn.to/3vBz5iu
Nikko Strom retweeted
Congrats to Amazon’s MiMi Aung 🎉 - just named to the 2021 @TIME 100 list for her leadership on Ingenuity, the @NASAJPL #MarsHelicopter. MiMi now works on Project Kuiper, helping bring affordable broadband to communities around the world. #TIME100 time.com/collection/100-most…
Nikko Strom retweeted
On the next episode of Alexa & Friends, Yang Liu, Alexa AI principal applied scientist, will be chatting with @jeffblankenburg about her work in speech recognition and understanding, prosody modeling, summarization, and NLP. Tune in on August 26 at 9am PT. amzn.to/3AVZjxo
Thanks @jeffblankenburg for the conversation – always fun to talk about ASR, AI tech, and about the future of Alexa. #AmazonScience youtube.com/watch?v=EJ5zjLOg…
Did you all know that a young grad student did e2e ASR in 1992? 😎 speech.kth.se/prod/publicati…
1
3
Alexa Conversations is our take on the future of building voice and multimodal customer experiences. See our latest paper here: #AmazonScience
amazon.science/publications/…
2
I's dotted, t's crossed. Done. Done. Done. nytimes.com/interactive/2020…
What’s next in voice tech? View video pitches from the class of Alexa Next Stage! Powered by @Techstars.
blog.aboutamazon.com/devices…
#AlexaFund #alexanextstage
2



