Do Smartphone Recordings Provide Quality Speech Data Good for Acoustic Analysis?
Mentor 1
Hanyong Park
Start Date
28-4-2023 12:00 AM
Description
An adequate quality recording is a prerequisite for conducting acoustic analysis of speech data. To obtain such a recording, researchers often collect data in a lab setting (e.g., a computer connected to a microphone using an audio interface). With this being an increasing difficulty due to COVID-19, we investigated whether smartphone recordings could be an alternative method to collect speech data for acoustic analysis. We brought ten native speakers of English to the UWM Phonetics Laboratory and recorded their readings of The North Wind and the Sun, once through lab recording equipment and again through their own smartphone using Awesome Voice Recorder, a lossless mobile phone application. Both recordings were conducted in a sound booth and the recording settings were the same in both conditions. To determine whether the phone recordings provide speech data as good as the lab recordings for acoustic analysis, we compared acoustic measures for monophthongs, which are vowel duration, F0 (i.e., pitch), F1, F2, and F3 (i.e., acoustic properties related to vowel quality), from the two recordings. If the acoustic measures were not different from each other, we would conclude that the phone recordings can be a viable option to collect data for phonetics research. We found that the acoustic measures were comparable across the two recordings for all measures except for F0. The phone recordings exhibited overall 5 Hz higher F0 than the lab recordings. The F0 difference, however, is negligible if we consider different pitch ranges among participants. Our results suggest that recordings using the participants’ own smartphones produce speech data well enough for acoustic analysis, at least for vowels, if recordings are conducted with a lossless mobile phone application and in a quiet environment.
Do Smartphone Recordings Provide Quality Speech Data Good for Acoustic Analysis?
An adequate quality recording is a prerequisite for conducting acoustic analysis of speech data. To obtain such a recording, researchers often collect data in a lab setting (e.g., a computer connected to a microphone using an audio interface). With this being an increasing difficulty due to COVID-19, we investigated whether smartphone recordings could be an alternative method to collect speech data for acoustic analysis. We brought ten native speakers of English to the UWM Phonetics Laboratory and recorded their readings of The North Wind and the Sun, once through lab recording equipment and again through their own smartphone using Awesome Voice Recorder, a lossless mobile phone application. Both recordings were conducted in a sound booth and the recording settings were the same in both conditions. To determine whether the phone recordings provide speech data as good as the lab recordings for acoustic analysis, we compared acoustic measures for monophthongs, which are vowel duration, F0 (i.e., pitch), F1, F2, and F3 (i.e., acoustic properties related to vowel quality), from the two recordings. If the acoustic measures were not different from each other, we would conclude that the phone recordings can be a viable option to collect data for phonetics research. We found that the acoustic measures were comparable across the two recordings for all measures except for F0. The phone recordings exhibited overall 5 Hz higher F0 than the lab recordings. The F0 difference, however, is negligible if we consider different pitch ranges among participants. Our results suggest that recordings using the participants’ own smartphones produce speech data well enough for acoustic analysis, at least for vowels, if recordings are conducted with a lossless mobile phone application and in a quiet environment.
