HARVARD speech corpus - audio recording 2019
Philippa Demonte · University of Salford Institutional Repository (University of Salford) · 2019
High-quality (sampling rate: 48 kHz; 32-bit rate) digital audio .wav files of a new recording of the HARVARD speech corpus in its entirety (720 phonetically balanced sentences), featuring a female native British English speaker. For use in speech-in-noise tests, evaluations of audio quality, machine learning, and so on. Recorded: December 2018 at the University of Salford. The audio files are licensed under the Creative Commons Attribution-NonCommercial 4.0 International License(https://creativecommons.org/licenses/by-nc/4.0/legalcode) Citation for the HARVARD sentences:* From the appendix of: 'IEEE Subcommittee on Subjective Measurements IEEE Recommended Practices for Speech Quality Measurements'. IEEE Transactions on Audio and Electroacoustics, vol. 17, pp. 227-246. 1969.* http://www-2.cs.cmu.edu/afs/cs.cmu.edu/project/fgdata/OldFiles/Recorder.app/utterances/Type1/harvsents.txt For an overview of the lists of HARVARD speech corpus sentences:* harvard.txt For an overview of how this version of the speech corpus was recorded and audio engineered:* harvard_201218_British_English_recording.txt For examples of the audio files at three different stages of processing* Speech corpus - example of raw audio: HARVARD list 01.wav* Speech corpus - example of edited audio: Harvard_L01_S01_0.wav* Speech corpus - example of edited end-pointed zero-padded audio: Harvard_L01_S01_5.wav For the .wav audio files of the HARVARD speech corpus in its entirety at different stages of processing, see the zip folders:* Speech corpus - Harvard - raw audio* Speech corpus - Harvard - edited and end-pointed audio* Speech corpus - Harvard - edited, end-pointed, zero-padded audio Each raw audio file is a recording of a single Harvard speech corpus list of 10 sentences.The two zip folders of the edited versions contain 10 individual sentence .wav files per sub-folder. For a .wav audio file of the room ambience in which this version of the Harvard speech corpus was recorded in:* Speech corpus - ambient noise of recording room For the Matlab 2018b scripts used for the end-pointing and zero-padding applied to the audio files:* EndPoint.m* zeropad.m