A new ElevenLabs collaboration with open-access science video platform ScienceCast enables anyone to generate instant narrated summaries of scientific papers.
Collaboration for Enhanced Access to Scientific Research
ElevenLabs’ capability for producing lifelike voiceovers now powers ScienceCast’s automatic video generation tool which condenses scientific research papers published on arXiv into 3 - 5 minute ‘elevator pitches’.
The integration is a step forward on all three companies’ commitment to making scientific content more accessible and promoting scientific collaboration through open access.
ScienceCast brings scientific research to a wider audience, including fellow researchers, journal editors, grant program managers, and others working at the cutting edge of science and innovation.
It adds community engagement and collaboration features to arXiv’s research-sharing platform as a member of arXiv Labs - a community framework for developing tools to enhance access to arXiv.
arXiv's Role and New Multimedia Features by ScienceCast
ArXiv provides an alternative to academic journals and peer review processes with a network of volunteer moderators who classify and validate research published on arXiv.org. It offers researchers broad range of services: article submission, compilation, production, retrieval, search and discovery.
More than two million academic articles have been published on arXiv.org across maths, science, computer science and engineering. In any given month the volume of papers published would take a person around 3 years to read.
ScienceCast’s feed of short videos and audio summaries (‘casts’) are directly linked to arXiv papers, allowing anyone to create a multimedia summary of an arXiv paper which can be shared elsewhere.
Innovative Voiceovers and the Vision for Open Science
With ElevenLabs’ functionality built into ScienceCasts’ video generation tool, video summaries are now automatically narrated by one of ElevenLabs’ ‘synthetic’ voices.
By including audio narration by default, ScienceCast content is now more accessible to visually impaired people and more digestible to researchers with limited time.
Voices used for ScienceCast videos were created from scratch by ElevenLabs, using generative AI rather than using a real person’s voice. For each paper summary, spoken audio content is generated through ElevenLabs’ speech synthesis functionality to narrate the video in this voice.
ScienceCast co-founder Victor Galitski said: "Our collaboration with ElevenLabs and arXiv is timely in the year of open science, a movement with goals that include increasing the accessibility of research, accelerating discovery, promoting trust and equity. We face a time when scientific data is overflowing at an accelerating pace, to the point where researchers struggle to keep up with the constant influx of papers. The joint effort between ScienceCast, ElevenLabs, and arXiv, provides a tangible example of using AI to offer an engaging and accessible alternative entry point into scientific results and data."
arXiv Scientific Director, Steinn Sigurdsson said: “The collaboration between ElevenLabs and ScienceCast is a major step forward to improving both access and accessibility for the latest research results. Together, we are providing an essential service to both researchers and members of society with interest in the latest developments in science and technology.”