Sony Pixel Power calrec Sony

How Deep Learning Is Aiding Preservation of Seneca and Other Endangered Languages

03/01/2019

Linguists estimate that at least half of the world's estimated 7,000 spoken languages will become extinct by the century's end, due to forces ranging from globalization to cultural assimilation.

Part of the challenge of documenting and revitalizing endangered languages is a lack of texts and speech recordings to work with. Seneca, a language of one of the six Iroquois Nations in North America, has only about 100 first-language speakers and several hundred more second-language learners.

Automatic speech recognition (ASR) technology is widely used to transcribe languages with millions or billions of speakers, like English and Mandarin. But it has only scratched the surface with languages like Seneca, which have vastly fewer speakers and significantly less data to work with.

Now a team of researchers at the Rochester Institute of Technology in New York, along with colleagues from the University at Buffalo, is tapping deep learning to bolster the ability of ASR. And while its focus is on Seneca, the project's vision encompasses the preservation of languages globally as well as an important part of our shared cultural history.

Knowing about different languages teaches us a lot about how our brain works, said Emily Prud'hommeaux, an assistant professor of computer science at Boston College and a research faculty member at RIT. When you document a language, you're preserving information not only about that language but also about how humans use language in general.

It's no coincidence that Prud'hommeaux and her team started with the Seneca language. Three members of the Seneca nation are part of the effort - a direct connection that is rare in research of this type, she said.

Leading the charge is Robbie Jimerson, a Ph.D. student in RIT's Golisano College of Computing and Information Science. He is a member of the Seneca Nation of Indians and is passionate about ensuring the survival of the Seneca language.

There's a big effort by the leaders of the tribe to preserve and promote our language, said Jimerson. I was looking for an opportunity to contribute.

Using GANs to Create More Language Samples Now in its third year, the project has had challenges when it comes to accumulating language data. Jimerson said the Seneca community can be guarded about what it shares with other people, so there wasn't an abundance of recordings of the language being spoken. He set out to change that.

He started by recording friends and elders who speak the language and asking them to record their friends. He found out whenever someone was speaking Seneca in public. He asked for family recordings of elders telling stories handed down from previous generations. And he grabbed any publicly available videos or recordings he could find online.

The team has fine-tuned an ASR model for Seneca, running it through generative adversarial networks to create more samples out of the limited number of recordings. The model turns wave files of the spoken language into streams of characters, while computing probability and making corrections.

The resulting data is fed into a deep learning model that in turn expands upon the ASR model's accuracy.

The team's networks run in two compute settings: on a nine-server machine learning lab running a variety of NVIDIA Tesla GPUs, and on a university cluster of large servers, each running 10 NVIDIA Tesla P4 GPUs. Each cluster runs a range of deep learning frameworks such as TensorFlow and Caffe.

The computer engineering cluster is for all students in the computer engineering department, and so they have to compete' for these resources, said Ray Ptucha, assistant professor of computer engineering at RIT, another collaborator on this project.

With access to these clusters at a premium, Jimerson tests code and checks the stability of models on a local machine running an NVIDIA TITAN X rather than inconvenience other students by running a model that might crash.

Achieving Better Accuracy So far, the team's efforts have brought the word error rate of its ASR model from 70 percent down to 56 percent. The goal, said Prud'hommeaux, is to get that rate down to 25 percent, which is where ASR systems were in processing English several years ago.

The more samples of spoken and written Seneca the team can accumulate, the more the error rate will decrease. (Today, English ASR models can achieve word error rates as low as 5 percent.)

The team's work is expected to help with language preservation efforts around the world.

Prud'hommeaux said the team has an agreement with an archiving institution that's a condition of a grant the project received from the National Science Foundation. The resulting language archiving database will be made available as a resource for other efforts seeking to document threatened languages.

Additionally, Prud'hommeaux said the team's work could prove helpful for any deep learning effort that has to make do with limited amounts of data.

Read more about the team's work in their research papers here and here.

Feature image: The Haudenosaunee (Iroquois Confederacy) flag, via Wikimedia Commons.
LINK: https://blogs.nvidia.com/blog/2019/01/02/deep-learning-preserves-senec...
See more stories from nvidia

Most recent headlines

09/11/2025

Dalet Unveils Agentic AI Media Workflows at IBC2025

Dalet today announced a transformative leap forward for media operations: Agentic Artificial Intelligence (AI) that unifies the Dalet ecosystem under one natura...

06/10/2025

France Tlvisions Wins Prestigious 2025 EBU Technology & Innovation Award in Groundbreaking Collaboration with Dalet

France T l visions, France's leading broadcaster, has received the 2025 EBU ...

13/09/2025

Cox Media Group's Misti Turnbull Inducted into the NATAS Silver Circle

ATLANTA Cox Media Group has announced that the company's vice president of news, Misty Turnbull has been inducted into the National Academy of Television Ar...

13/09/2025

Shotoku Debuts Swoop Cranes for Studio Robotics at IBC2025

AMSTERDAM Shotoku Broadcast Systems, a major developer of robotic systems, has announced plans to take studio robotics to the next level at IBC2025 by debuting ...

13/09/2025

Riedel Unveils Ultra-Light Bolero Mini Wireless Intercom...

At IBC2025 in Amsterdam, Riedel Communications unveiled Bolero Mini, the company's lightest and flattest wireless intercom beltpack to date. Designed to del...

13/09/2025

Shotoku Takes Studio Robotics to New Heights with IBC Deb...

Shotoku Broadcast Systems, the international developer of dependable, userfriendly robotic systems, is taking studio robotics to the next level at IBC 2025 with...

13/09/2025

The Bitmovin Video Developer Report 2025-26 Reveals Cost...

Bitmovin, a leading provider of video streaming solutions, today released the 9th annual Video Developer Report 2025/26, offering an in-depth look at the evolvi...

13/09/2025

Bitmovin and StreamShark Partner to Deliver High Quality...

Bitmovin, the leading provider of video streaming solutions, today announced a strategic partnership with StreamShark, the trusted video platform for enterprise...

13/09/2025

Ikegami Announces VFE-P711AD 7-inch OLED Multiformat On-C...

Ikegami has chosen IBC 2025 in Amsterdam as the launch venue for a major addition to its range of viewfinders. The new VFE-P711AD is a 7-inch high resolution OL...

13/09/2025

KitBash3D and Greyscalegorilla Announce Merger

Founder-led Merger to Fast Track R&D, Asset Library Upgrades, Tools and More; No Disruption to Pricing or Support for Users Today, KitBash3D, a pioneer in 3D a...

13/09/2025

Mavis Puts Itself at the Heart of Mobile Production

With NDI certification, Atomos integration, Grass Valley collaboration, and a new Monitor app, at this year's IBC, Mavis is showcasing a series of powerful...

13/09/2025

Creamsource Expands Vortex Family with Vortex24 Soft

Creamsource, maker of artisan LED lighting for film and television, has unveiled the Vortex24 Soft (V24S), a 1950W native soft light and the largest soft source...

13/09/2025

DAZN streams 2025 FIFA Club World Cup to billions of fans...

When international sports streaming service DAZN secured the global rights to the 2025 FIFA Club World Cup football tournament, it set out to deliver an unmatch...

13/09/2025

Riedel Communications Acquires hi human interface

Riedel Communications today announced the acquisition of hi human interface from Broadcast Solutions, bringing a powerful, vendor-agnostic control system to it...

13/09/2025

RTW chooses Calrec as technology partner for its AI ready...

Building on its long-term relationship with audio metering specialist RTW, Calrec has integrated the company's brand new TMxCore metering platform across it...

13/09/2025

Calrec unveils 48 fader Argo M at IBC2025 and demonstrate...

Calrec is expanding its family of future-ready self-contained Argo M control surfaces at IBC2025, with the addition of a brand new powerful 48-fader console. Co...

13/09/2025

September 11, 2025

Researchers find brain region that fuels compulsive drinking Study by Scripps Research scientists shows how the brain learns to seek alcohol for relief, not jus...

12/09/2025

College Football Kickoff 2025: Fox Sports Ups Look as Canon, Sony Power Shallow Focus Coverage

College Football Kickoff 2025: Fox Sports Ups Look as Canon, Sony Power Shallow ...

12/09/2025

ABC/ESPN Excited For WNBA Postseason Coverage In Revamped Format

ABC/ESPN Excited For WNBA Postseason Coverage In Revamped FormatThe Finals moves to a best-of-seven series in 2025By Mark J Burns, SVG Contributor Friday, Sep...

12/09/2025

Rabbit Trap Pulsates With Folklore Dread

(L-R) Jade Croot, Rosy McEwen, and Bryn Chainey attend the 2025 Sundance Film Festival premiere of Rabbit Trap at Eccles Theatre on January 24, 2025, in Park ...

12/09/2025

Spotify's The Drop Weekly' Brings You the Week in New Releases, Straight From Our Editors

For fans, we know how important it is to stay plugged into music culture and dis...

12/09/2025

Agama and Consult Red announce RDK Accelerator integration

Link ping, Sweden and Shipley, United Kingdom, September 12, 2025 - Agama, the expert in video observability and analytics for service quality and customer expe...

12/09/2025

IBC2025 Opens for Business

IBC2025 began on Sept. 12, with exhibits and conferences running through Sept. 15 at the RAI Amsterdam Convention Center. Explore the full TV Tech coverage of t...

12/09/2025

The Best Fictional Bands (and the Artists Who Make Them Great)

The Best Fictional Bands (and the Artists Who Make Them Great) With Spinal Tap II: The End Continues hitting theaters and songs from KPop Demon Hunters ruling...

12/09/2025

Tom Baldassare Joins Advanced Systems Group

Industry veteran Tom Baldassare has joined Advanced Systems Group, LLC (ASG), a technology and services provider for media creatives and content owners, as a Se...

12/09/2025

Maxon Unveils a Brand New Look for its Growing Family of...

Maxon, maker of powerful, approachable software solutions for creators working in 2D and 3D design, motion graphics, visual effects, and more, today announced a...

12/09/2025

PlayBox Neo US Partners with AI-Media to Deliver Scalable...

PlayBox Neo, a leading provider of media playout solutions, has partnered with AI-Media, pioneering developers of AI-powered captioning technology, to integrate...

12/09/2025

Dalet Unveils Agentic AI Media Workflows at IBC2025

Dalet today announced a transformative leap forward for media operations: Agentic Artificial Intelligence (AI) that unifies the Dalet ecosystem under one natura...

12/09/2025

Keepit and Ingram Micro launch strategic sales agreement...

New alliance strengthens the IT channel in Germany and Switzerland in protecting business-critical SaaS data. Keepit, the world s only independent, cloud-nativ...

12/09/2025

Mediaset selects Fincons Group AllRights to evolve rights...

Fincons Group, an international IT business consultancy and systems integrator company with more than 40 years of experience in the market, is proud to announce...

12/09/2025

EVS Acquires XD motion

Following its acquisition of Telemetrics, EVS continues its push into robotics with an announcement at IBC2025 that it is acquiring XD motion....

12/09/2025

Televisa Executive Joins NABA Board

TORONTO The North American Broadcasters Association (NABA) has announced the appointment of Eduardo Ruiz Sanchez, deputy director, broadcast operations at Telev...

12/09/2025

Ed Miller, Former SBE President, Has Died

Ed Miller, a longtime broadcast engineer in Ohio and a former national president of the Society of Broadcast Engineers, has died....

12/09/2025

IBC2025: Dynamic HDR Gains Traction

AMSTERDAM At this year's IBC2025, the Advanced HDR by Technicolor initiative will be pushing broadcasters to adopt a more dynamic, frame-by-frame conversion...

12/09/2025

Granville opens up one last time for U&GOLD in Open All Hours: Inside Out

Feature-length retrospective from Studio Crook to air in 2026 Sir David Jason returns to the nation's favourite comedy channel, U&GOLD, for Open All Hours:...

12/09/2025

Bob Geldof to receive Lifetime Achievement Award at the Sky Arts Awards 2025

Friday 12 September 2025 The Boomtown Rats, Nyah Grace, Soweto Kinch, Royal Ballet and Madness also announced to perform at the ceremony on Tuesday Sky today ...

12/09/2025

Riedel Unveils Ultra-Light Bolero Mini Wireless Intercom Beltpack

Wuppertal September 12, 2025 Riedel Unveils Ultra-Light Bolero Mini Wireless Intercom BeltpackAt IBC2025 in Amsterdam, Riedel Communications unveiled Bolero M...

12/09/2025

Riedel Communications Acquires hi human interface

Wuppertal September 12, 2025 Riedel Communications Acquires hi human interfaceRiedel Communications today announced the acquisition of hi human interface fro...

12/09/2025

New International Crime Series Road (WT)' Explores Twisted Murders Across Borders

Back to All News New International Crime Series Road (WT)' Explores Twiste...

12/09/2025

First Look: Thai Crime Drama Everybody Loves Me When I'm Dead' Premieres October 14

Back to All News First Look: Thai Crime Drama Everybody Loves Me When I'm ...

12/09/2025

Netflix Marks 10 Years in Japan, Announces Three New Series That Will Keep You Hitting The Next Episode

Back to All News Netflix Marks 10 Years in Japan, Announces Three New Series Th...

12/09/2025

What Is CORE+ Technologyand How Does It Elevate Church Sound?

CORE+ virtually removes distortion, setting a new standard for church sound and giving worship teams the clarity and confidence they need. Read the full artic...

12/09/2025

Margot Robbie, Colin Farrell, Mary Robinson and Conor Murray amongst guests on Late Late Show season opener

The Late Late Show is back with a bang after the summer break, and Patrick Kielt...

12/09/2025

Another jam-packed weekend of live, free-to-air Sport across RT

The World Athletics Championships, Ireland v France in the Women's Rugby World Cup quarter-final, the Irish Champions Festival, and two Sports Direct Men...

12/09/2025

Katie Hannon explores the shelves of Ireland's National Archives in new series

The Records Show starts Sunday at 6.30pm on RT One and RT Player. Katie Hanno...

11/09/2025

Report: Busy Live Sports Streaming Execs Have Low-hanging Fruit' in Front of Them

Report: Busy Live Sports Streaming Execs Have Low-hanging Fruit' in Front o...

11/09/2025

Inside Game Creek Video's Big Week as Ovation, Flagship Make NFL Debuts

Inside Game Creek Video's Big Week as Ovation, Flagship Make NFL DebutsBy Ken Kerschbaumer, Editorial Director Thursday, September 11, 2025 - 7:00 am Pr...

11/09/2025

NFL Kickoff 2025: Prime Sports Starts New Season at Lambeau Field; Sets Sights on Holiday Matchups, Second-Ever Playoff Game

NFL Kickoff 2025: Prime Sports Starts New Season at Lambeau Field; Sets Sights o...

11/09/2025

College Football Kickoff 2025: NBC Sports Pushes HDR Image Quality, Aerial Drones for Big Ten Saturday Night' and Notre Dame on NBC'

College Football Kickoff 2025: NBC Sports Pushes HDR Image Quality, Aerial Drone...