Fearless Steps: Apollo-11 Corpus Advancements for Speech Technologies from Earth to the Moon

Abstract

The Apollo Program is one of the most significant benchmarks for technology and innovation in human history. The previously introduced UTD-CRSS Fearless Steps initiative resulted in the digitization of the original analog audio tapes recorded during the Apollo Space Missions. The entire speech data for the Apollo 11 Mission is now being made publicly available with the release of the Fearless Steps Corpus. This corpus consists of a cumulative 19,000 hours of conversational speech spanning over thirty time-synchronized channels. With over six hundred speakers, the corpus has a rich collection of information which can be beneficial for research and advancement in the speech and language community. Recent efforts on this data have led to the generation of pipeline diarization transcripts for the entire speech corpus. Research has also been done to address speech and natural language tasks such as speech activity detection, speech recognition, and sentiment analysis. This paper provides an overview of the Fearless Steps Corpus and highlights the factors that make the processing of this data a challenging problem. To promote further development of algorithms for naturalistic data, five challenge tasks are also organized. We also describe the challenge tasks with details on a fully transcribed subset of the corpus, and initial results achieved by our systems.

The 2019 Inaugural Fearless Steps Challenge: A Giant Leap for Naturalistic Audio

“…To ensure an equitable distribution of data into training, evaluation, and development sets for the challenge tasks, we have categorized the data based on noise levels, amount of speech content, and amount of silence. Due to the long silence durations for some channels, and based on importance of the mission, the speech activity density of the corpus varies throughout the mission [1]. A total of 80 hours of audio are provided for task system development.…”

“…The last six years have seen the development of the Corpus consisting of over 19,000 hours of audio data from the Apollo 1, 11, 13 and the Gemini 8 missions. The unique nature of this data posed a serious challenge for analysis using conventional speech technologies [1]. This challenge motivated the development of multiple solutions from CRSS catered to the nature and complexity of the Apollo data [2,3,4,5,6,7].…”

Multi-Level Adaptive Speech Activity Detector for Speech in Naturalistic Environments

“…The Fearless Steps Challenge 1 2019 is organized to evaluate the state-of-the-art methods for different applications with naturalistic audio signals in challenging environments [29,30]. A corpus from the data captured during Apollo-11 Mission is released for the challenge.…”

“…The combo SAD has been proposed by the challenge organizers in their previous investigations [33]. The speech recorded in the corpus are unprompted, and hence subject to significant variations in speech characteristics for every speaker [29].…”

Speaker Recognition Benchmark Using the CHiME-5 Corpus

“…Finally, the currently ongoing VOiCES [7] and Fearless Steps [8] challenges also explore interesting areas. VOiCES focuses on robustness to reverberation and background noises of replayed speech, while Fearless Steps defines a more holistic set of challenges using data from the Apollo-11 mission.…”