● Turnaround times are slow and cost is high for human-generated transcription services.
● Out-of-the-box automatic speech-to-text solutions are not accurate enough.
● Training speech-to-text software is complicated work that few organizations are likely to undertake, given the significant time and resources required.
That’s where we come in. By addressing needs common to news organizations and audio archives, Pop Up Archive is bringing cutting edge speech recognition methods to organizations that would otherwise never benefit from this technology.
We’re offering an exclusive pilot opportunity to U.S. media organizations and archives seeking customized speech-to-text software for use in transcription and captioning.
● The eight week pilot will create unique speech-to-text vocabularies specifically tailored to these organizations, built directly from words and phrases commonly found in their content.
● The resulting software will be entirely automated and will effectively gauge likeliness that sounds in a variety of contexts (including poor quality audio and background noise) correspond to words or phrases common to pilot organizations’ content.
Our software was initially trained on a subset of audio and transcripts from PBS, the Broadcast Board of Governors, the Washington Post, and numerous independent producers, reporters, and radio stations.
Advised by the British Broadcasting Corp. R&D team and partnered with the Public Radio Exchange, Pop Up Archive is supported by the Knight Foundation, the National Endowment for the Humanities, and 500 Startups.