Martí Umbert i Morist (@martiumbert) header
Martí Umbert i Morist (@martiumbert) profile picture

Senior Researcher in the Language Technologies Lab at @BSC_CNS Previously: @aflorithmic @verbiocorporate @telefonicaid & Yamaha Corp Japan / PhD @mtg_upf

Barcelona
Joined May 2011
Photo by Martí Umbert i Morist (@martiumbert)Photo by Martí Umbert i Morist (@martiumbert)Photo by Martí Umbert i Morist (@martiumbert)Photo by Martí Umbert i Morist (@martiumbert)Photo by Martí Umbert i Morist (@martiumbert)Photo by Martí Umbert i Morist (@martiumbert)Photo by Martí Umbert i Morist (@martiumbert)Photo by Martí Umbert i Morist (@martiumbert)Photo by Martí Umbert i Morist (@martiumbert)Photo by Martí Umbert i Morist (@martiumbert)
Martí Umbert i Morist retweeted
Ariadna Font, excap de l'àrea d'ètica de Twitter: "X és una caricatura del que podria haver estat" 3cat.cat/324/ariadna-font-ex…
86
13
24
88
37,354
🌟 Discover AI's potential and challenges in @derspiegel's latest cover story. @aflorithmic's Co-Founder Björn Ühss shares thoughts on AI-voice cloning and the evolving quality of synthetic voices. 🚀 Read the article here: 🔗spiegel.de/netzwelt/web/kuen…
1
2
214
Martí Umbert i Morist retweeted
Generative AI Landscape Happy to see a lot of companies I have personally worked it. Kudos to the teams building the future of AI. Source: @Base10Partners Bookmark this for reference 📌
35
148
17
739
Martí Umbert i Morist retweeted
@aflorithmic was ranked in the Top 101 #Audio companies in London 🎧. We strive everyday to provide our customers with the most powerful Audio-As-A-Service platform for building #AI audio at scale 🚀 Read more 🔗: bit.ly/topaudiostartups #aflorithmic #startups #innovation #tech
2
5
Martí Umbert i Morist retweeted
What would happen if #customers were at the center of your #business, and what would happen if those customers could always be served #automatically? What if they could do all that with their #voice autonomously? lnkd.in/eYxBevu
1
2
Martí Umbert i Morist retweeted
The MTG is looking to hire a researcher to work on the development of machine learning algorithms for music audio analysis and representation for music recommendation systems, with a further possibility of joining the PhD program on these topics. upf.edu/web/mtg/home/-/asset…
Martí Umbert i Morist retweeted
No matter the language, we exchange information at 39 bits/second, suggesting a biological limit. Languages that are lower information density are spoken fast (Spanish & Japanese) while denser languages are spoken more slowly (Thai, Vietnamese, Chinese). advances.sciencemag.org/cont…
55
1,014
279
3,668
Martí Umbert i Morist retweeted
A semi-supervised multi-head approach to learning speech quality scores can yield over 35% improvement wrt SOTA! Check SESQA, our latest work, at arxiv.org/abs/2010.00368 Work with @jordiponsdotme and @santty128 done at @Dolby AI.
3
18
5
88
Martí Umbert i Morist retweeted
How do you safely verify the identity of employees or customers through the unique power of their #Voice? Learn why the Canada Border Services Agency chose Verbio’s #VoiceBiometrics technology to improve their customs services and security measures: bit.ly/2AjKIm8
2
4
Martí Umbert i Morist retweeted
Artie releases tool to measure bias in speech recognition models ift.tt/2Os2mHx
Martí Umbert i Morist retweeted
👋 New Video about @DeepMind s new End-to-End Adversarial TTS system. Turns out you can overcome the alignment problem via smart dynamic programming! youtu.be/WTB2p4bqtXU @sedielem @MikoYBinkowski @erich_elsen
1
17
125
Martí Umbert i Morist retweeted
We are happy to announce the release of a new MTG video, a short presentation that highlights the most relevant aspects of our work. Read our blog post to know more about the process of creation of the video and what is behind it: upf.edu/web/mtg/blog/-/asset… We hope you like it!
1
22
2
64
Martí Umbert i Morist retweeted
JOIN US! We are looking to hire a researcher/developer to start as soon as possible. More details here: upf.edu/web/mtg/home/-/asset…
Martí Umbert i Morist retweeted
We present TensorFlow audio models in Essentia! New algorithms and pre-trained models for deep learning inference in your C++ and Python applications. Blog: mtg.github.io/essentia-labs/… Paper at #ICASSP2020: arxiv.org/abs/2003.07393 Contact: @pablo__alonso & @di_bogdanov
1
24
3
44
Martí Umbert i Morist retweeted
We are presenting at #ICASSP2020 a demo of our work on Choir Singing Synthesis in collaboration with @mtg_upf. Video presentation is now online. We show our prototype for choir singing rehearsal in the scope of the @TrompaMusic project. 2020.ieeeicassp-virtual.org/… @voiceful_io
1
27
55
Martí Umbert i Morist retweeted
Introducing Jukebox, a neural net that generates music, including rudimentary singing, as raw audio in a variety of genres and artist styles. We're releasing a tool for everyone to explore the generated samples, as well as the model and code: openai.com/blog/jukebox/
201
2,138
751
7,489
Martí Umbert i Morist retweeted
Finding it hard to get through the volume of customer calls? ☎️ Let our vailable.ai take care of them 📞 Introducing our new Cloud Solution to keep customers cared for 24/7, with a sophisticated Voice AI service🤗 Learn more: web.vailable.ai/hello/ #customerservice
Martí Umbert i Morist retweeted
Freesound - "Happy 15th birthday Freesound!" by MTG freesound.org/people/MTG/sou…
6
15