In a paper originally published last October and accepted to the International Conference on Learning Representations (ICLR) 2020, researchers affiliated with Google and the University College London propose an AI model that enables control of speech characteristics like pitch, emotion, and speaking rate with as little as 30 minutes of data.
The work has obvious commercial implications. Brand voices such as Progressives Flo (played by comedian Stephanie Courtney) are often pulled in for pick-ups sessions to address mistakes, changes, or additions in voiceover scripts long after a recording finishes. AI-assisted voice correction could eliminate the need for these, saving time and money on the part of the actors employers.
A previous study investigated the use of so-called style tokens (which represented different categories of emotion) to control speech affect. The method achieved good results with only 5% of labeled data, but it couldnt handle speech samples with varying prosody (i.e., intonation, tone, stress, and rhythm) and fixed emotion. The work from Google and the University of College London addresses this limitation.
The researchers trained the system for 300,000 steps across 32 of Googles custom-designed tensor processing units (TPUs), a scale of compute exceeding that used in previous work. They report that using 30 minutes of labeled data allowed for a significant degree of control over speech rate, valence, and arousal, and that affect accuracy didnt degrade noticeably with at least 10% of labeled data. The researchers said that just 3 minutes of data allowed for control of speech rate and extrapolation outside data seen during training a result the researchers claim beat out state-of-the-art baselines.
The researchers system taps a trained generative model that can synthesize acoustic features from text. Similar to Googles Tacotron 2, a text-to-speech (TTS) system that generates natural-sounding speech from raw transcripts, the new system can produce visual representations of frequencies called spectrograms by training a second model such as DeepMinds WaveNet to act as a vocoder, a voice codec that analyzes and synthesizes voice data. (This system uses WaveRNN.)
An annotated data set comprising 72,405 roughly 5-second recordings from 40 English speakers, amounting to 45 hours of audio, was used to train the system. The speakers, all of whom were trained voice actors, were prompted to read text snippets with varying levels of valence (emotions like sadness or happiness) and arousal (excitement or energy). From these sessions, the researchers obtained six possible affective states, which they modeled and use as labels along with labels for speaking rate (here defined as the number of syllables per second in each utterance).
Heres one of the voices the system modified (which sounds not unlike the default Google Assistant voice, interestingly) to have high arousal and an angry valence:
And heres that same voice with high arousal and a happy valence:
And low arousal and sad valence:
The studys coauthors acknowledge that the work might raise ethical concerns because it could be misused for misinformation or to commit fraud. Indeed, deepfakes media that takes a person in an existing image, audio recording, or video and replaces them with someone elses likeness using AI are multiplying quickly, and have already been used to defraud a major energy producer. In tandem with tools like Resemble, Baidus Deep Voice, and Lyrebird, which need only seconds to minutes of audio samples to clone someones voice, its not difficult to imagine how this new system might add fuel to the fire.
But the coauthors also assert that in this case, since the focus of this work is on improved prosody with potential benefits to human-computer interfaces, the benefits likely outweigh the risks. We urge the research community to take seriously the potential for misuse both of this work and broader advances in TTS, they wrote.
See the article here:
Googles AI can adjust voice emotion, pitch, and speed with 30 minutes of data - VentureBeat
- Working at DeepMind | Glassdoor [Last Updated On: September 8th, 2019] [Originally Added On: September 8th, 2019]
- DeepMind Q&A Dataset - New York University [Last Updated On: October 6th, 2019] [Originally Added On: October 6th, 2019]
- Google absorbs DeepMind healthcare unit 10 months after ... [Last Updated On: October 7th, 2019] [Originally Added On: October 7th, 2019]
- deep mind Mathematics, Machine Learning & Computer Science [Last Updated On: November 1st, 2019] [Originally Added On: November 1st, 2019]
- Health strategies of Google, Amazon, Apple, and Microsoft - Business Insider [Last Updated On: November 21st, 2019] [Originally Added On: November 21st, 2019]
- To Understand The Future of AI, Study Its Past - Forbes [Last Updated On: November 21st, 2019] [Originally Added On: November 21st, 2019]
- Tremor patients can be relieved of the shakes for THREE YEARS after having ultrasound waves - Herald Publicist [Last Updated On: November 21st, 2019] [Originally Added On: November 21st, 2019]
- The San Francisco Gay Mens Chorus Toured the Deep South - SF Weekly [Last Updated On: November 21st, 2019] [Originally Added On: November 21st, 2019]
- The Universe Speaks in Numbers: The deep relationship between math and physics - The Huntington News [Last Updated On: November 21st, 2019] [Originally Added On: November 21st, 2019]
- MINI John Cooper Works GP is a two-seater hot hatch that shouts its 306 HP - SlashGear [Last Updated On: November 21st, 2019] [Originally Added On: November 21st, 2019]
- How To Face An Anxiety Provoking Situation Like A Champion - Forbes [Last Updated On: November 21st, 2019] [Originally Added On: November 21st, 2019]
- The Most Iconic Tech Innovations of the 2010s - PCMag [Last Updated On: November 24th, 2019] [Originally Added On: November 24th, 2019]
- Why tech companies need to hire philosophers - Quartz [Last Updated On: November 24th, 2019] [Originally Added On: November 24th, 2019]
- Living on Purpose: Being thankful is a state of mind - Chattanooga Times Free Press [Last Updated On: November 24th, 2019] [Originally Added On: November 24th, 2019]
- EDITORIAL: West explosion victims out of sight and clearly out of mind - Waco Tribune-Herald [Last Updated On: November 24th, 2019] [Originally Added On: November 24th, 2019]
- Do you need to sit still to be mindful? - The Sydney Morning Herald [Last Updated On: November 26th, 2019] [Originally Added On: November 26th, 2019]
- Listen To Two Neck Deep B-Sides, Beautiful Madness And Worth It - Kerrang! [Last Updated On: November 26th, 2019] [Originally Added On: November 26th, 2019]
- Worlds Last Male Northern White Rhino Brought Back To Life Using AI - International Business Times [Last Updated On: November 26th, 2019] [Originally Added On: November 26th, 2019]
- Eat, drink, and be merryonly if you keep in mind these food safety tips - Williamsburg Yorktown Daily [Last Updated On: November 26th, 2019] [Originally Added On: November 26th, 2019]
- The alarming trip that changed Jeremy Clarksons mind on climate change - The Week UK [Last Updated On: November 26th, 2019] [Originally Added On: November 26th, 2019]
- Actionable Insights on Artificial Intelligence in Law Market with Future Growth Prospects by 2026 | AIBrain, Amazon, Anki, CloudMinds, Deepmind,... [Last Updated On: November 26th, 2019] [Originally Added On: November 26th, 2019]
- Searching for the Ghost Orchids of the Everglades - Discover Magazine [Last Updated On: November 26th, 2019] [Originally Added On: November 26th, 2019]
- Parkinsons tremors could be treated with SOUNDWAVES, claim scientists - Herald Publicist [Last Updated On: November 26th, 2019] [Originally Added On: November 26th, 2019]
- Golden State Warriors still have prolonged success in mind - Blue Man Hoop [Last Updated On: November 26th, 2019] [Originally Added On: November 26th, 2019]
- 3 Gratitude Habits You Can Adopt Over The Thanksgiving Holiday For Deeper Connection And Joy - Forbes [Last Updated On: November 26th, 2019] [Originally Added On: November 26th, 2019]
- The minds that built AI and the writer who adored them. - Mash Viral [Last Updated On: November 28th, 2019] [Originally Added On: November 28th, 2019]
- Parkinson's Patients are Mysteriously Losing the Ability to Swim After Treatment - Discover Magazine [Last Updated On: November 28th, 2019] [Originally Added On: November 28th, 2019]
- Hannah Fry, the woman making maths cool | Times2 - The Times [Last Updated On: November 28th, 2019] [Originally Added On: November 28th, 2019]
- Meditate with Urmila: Find balance of body, mind and breath - Gulf News [Last Updated On: November 28th, 2019] [Originally Added On: November 28th, 2019]
- We have some important food safety tips to keep in mind while cooking this Thanksgiving - WQOW TV News 18 [Last Updated On: November 28th, 2019] [Originally Added On: November 28th, 2019]
- Being thankful is a state of mind | Opinion - Athens Daily Review [Last Updated On: December 2nd, 2019] [Originally Added On: December 2nd, 2019]
- Can Synthetic Biology Inspire The Next Wave of AI? - SynBioBeta [Last Updated On: December 2nd, 2019] [Originally Added On: December 2nd, 2019]
- LIVING ON PURPOSE: Being thankful is a state of mind - Times Tribune of Corbin [Last Updated On: December 2nd, 2019] [Originally Added On: December 2nd, 2019]
- AI Hardware Summit Europe launches in Munich, Germany on 10-11 March 2020, the ecosystem event for AI hardware acceleration in Europe - Yahoo Finance [Last Updated On: December 5th, 2019] [Originally Added On: December 5th, 2019]
- Of course Facebook and Google want to solve social problems. Theyre hungry for our data - The Guardian [Last Updated On: December 5th, 2019] [Originally Added On: December 5th, 2019]
- Larry, Sergey, and the Mixed Legacy of Google-Turned-Alphabet - WIRED [Last Updated On: December 6th, 2019] [Originally Added On: December 6th, 2019]
- AI Index 2019 assesses global AI research, investment, and impact - VentureBeat [Last Updated On: December 11th, 2019] [Originally Added On: December 11th, 2019]
- For the Holidays, the Gift of Self-Care - The New York Times [Last Updated On: December 11th, 2019] [Originally Added On: December 11th, 2019]
- Stopping a Mars mission from messing with the mind - Axios [Last Updated On: December 11th, 2019] [Originally Added On: December 11th, 2019]
- Feldman: Impeachment articles are 'high crimes' Founders had in mind | TheHill - The Hill [Last Updated On: December 11th, 2019] [Originally Added On: December 11th, 2019]
- Opinion | Frankenstein monsters will not be taking our jobs anytime soon - Livemint [Last Updated On: December 11th, 2019] [Originally Added On: December 11th, 2019]
- DeepMind co-founder moves to Google as the AI lab positions itself for the future - The Verge [Last Updated On: December 11th, 2019] [Originally Added On: December 11th, 2019]
- Google Isn't Looking To Revolutionize Health Care, It Just Wants To Improve On The Status Quo - Newsweek [Last Updated On: December 12th, 2019] [Originally Added On: December 12th, 2019]
- Artificial Intelligence Job Demand Could Live Up to Hype - Dice Insights [Last Updated On: December 12th, 2019] [Originally Added On: December 12th, 2019]
- What Are Normalising Flows And Why Should We Care - Analytics India Magazine [Last Updated On: December 15th, 2019] [Originally Added On: December 15th, 2019]
- Terence Crawford has next foe in mind after impressive knockout win - New York Post [Last Updated On: December 15th, 2019] [Originally Added On: December 15th, 2019]
- DeepMind proposes novel way to train safe reinforcement learning AI - VentureBeat [Last Updated On: December 15th, 2019] [Originally Added On: December 15th, 2019]
- Winning the War Against Thinking - So you've emptied your brain. Now what? - Chabad.org [Last Updated On: December 16th, 2019] [Originally Added On: December 16th, 2019]
- 'Echo Chamber' as Author of the 'Hive Mind' - Ricochet.com [Last Updated On: December 16th, 2019] [Originally Added On: December 16th, 2019]
- Lindsey Graham: 'I Have Made Up My Mind' to Exonerate Trump and 'Don't Need Any Witnesses' WATCH - Towleroad [Last Updated On: December 16th, 2019] [Originally Added On: December 16th, 2019]
- Blockchain in Healthcare Market to 2027 By Top Leading Players: iSolve LLC, Healthcoin, Deepmind Health, IBM Corporation, Microsoft Corporation,... [Last Updated On: December 16th, 2019] [Originally Added On: December 16th, 2019]
- In sight but out of mind - The Hindu [Last Updated On: December 16th, 2019] [Originally Added On: December 16th, 2019]
- The Case for Limitlessness Has Its Limits: Review of Limitless Mind by Joe Boaler - Education Next - EducationNext [Last Updated On: December 16th, 2019] [Originally Added On: December 16th, 2019]
- The Top 10 Diners In Deep East Texas, According To Yelp - ksfa860.com [Last Updated On: December 16th, 2019] [Originally Added On: December 16th, 2019]
- 3 breathing exercises to reduce stress, anxiety and a racing mind - Irish Examiner [Last Updated On: December 16th, 2019] [Originally Added On: December 16th, 2019]
- DeepMind exec Andrew Eland leaves to launch startup - Sifted [Last Updated On: December 16th, 2019] [Originally Added On: December 16th, 2019]
- The Top 10 Diners In Deep East Texas, According To Yelp - kicks105.com [Last Updated On: December 17th, 2019] [Originally Added On: December 17th, 2019]
- Mind the Performance Gap New Future Purchasing Category Management Report Out Now - Spend Matters [Last Updated On: December 17th, 2019] [Originally Added On: December 17th, 2019]
- Madison singles and deep cuts that stood out in 2019 - tonemadison.com [Last Updated On: December 19th, 2019] [Originally Added On: December 19th, 2019]
- Hilde Lee: Latkes bring an ancient miracle to mind on first night of Hanukkah - The Daily Progress [Last Updated On: December 19th, 2019] [Originally Added On: December 19th, 2019]
- Political Cornflakes: Trump responds to impeachment with complaints about the 'deep state' and toilet flushing - Salt Lake Tribune [Last Updated On: December 19th, 2019] [Originally Added On: December 19th, 2019]
- Google CEO Sundar Pichai Is the Most Expensive Tech CEO to Keep Around - Observer [Last Updated On: December 24th, 2019] [Originally Added On: December 24th, 2019]
- Christmas Lectures presenter Dr Hannah Fry on pigeons, AI and the awesome power of maths - inews [Last Updated On: December 24th, 2019] [Originally Added On: December 24th, 2019]
- The ultimate guitar tuning guide: expand your mind with these advanced tuning techniques - Guitar World [Last Updated On: December 24th, 2019] [Originally Added On: December 24th, 2019]
- Inside The Political Mind Of Jerry Brown - Radio Ink [Last Updated On: December 24th, 2019] [Originally Added On: December 24th, 2019]
- Elon Musk Fact-Checked His Own Wikipedia Page and Requested Edits Including the Fact He Does 'Zero Investing' - Entrepreneur [Last Updated On: December 24th, 2019] [Originally Added On: December 24th, 2019]
- The 9 Best Blobs of 2019 - Livescience.com [Last Updated On: December 24th, 2019] [Originally Added On: December 24th, 2019]
- AI from Google is helping identify animals deep in the rainforest - Euronews [Last Updated On: December 24th, 2019] [Originally Added On: December 24th, 2019]
- Want to dive into the lucrative world of deep learning? Take this $29 class. - Mashable [Last Updated On: December 24th, 2019] [Originally Added On: December 24th, 2019]
- Re: Your Account Is Overdrawn - Thrive Global [Last Updated On: December 27th, 2019] [Originally Added On: December 27th, 2019]
- Review: In the Vale is full of characters who linger long in the mind - Nation.Cymru [Last Updated On: December 27th, 2019] [Originally Added On: December 27th, 2019]
- 10 Gifts That Cater to Your Loved One's Basic Senses - Wide Open Country [Last Updated On: December 27th, 2019] [Originally Added On: December 27th, 2019]
- The Most Mind-Boggling Scientific Discoveries Of 2019 Include The First Image Of A Black Hole, A Giant Squid Sighting, And An Exoplanet With Water... [Last Updated On: December 27th, 2019] [Originally Added On: December 27th, 2019]
- DeepMind's new AI can spot breast cancer just as well as your doctor - Wired.co.uk [Last Updated On: January 1st, 2020] [Originally Added On: January 1st, 2020]
- Why the algorithms assisting medics is good for health services (Includes interview) - Digital Journal [Last Updated On: January 4th, 2020] [Originally Added On: January 4th, 2020]
- 2020: The Rise of AI in the Enterprise - IT World Canada [Last Updated On: January 4th, 2020] [Originally Added On: January 4th, 2020]
- An instant 2nd opinion: Google's DeepMind AI bests doctors at breast cancer screening - FierceBiotech [Last Updated On: January 4th, 2020] [Originally Added On: January 4th, 2020]
- Google's DeepMind AI outperforms doctors in identifying breast cancer from X-ray images - Business Insider UK [Last Updated On: January 4th, 2020] [Originally Added On: January 4th, 2020]
- New AI toolkit from the World Economic Forum is promising because it's free - The National [Last Updated On: January 20th, 2020] [Originally Added On: January 20th, 2020]
- AKA Wants to Help People Break Bad Habits and Create New Positive Ones - Hospitality Net [Last Updated On: January 20th, 2020] [Originally Added On: January 20th, 2020]