We compare various USB and LENA devices that could be used to collect long-form-recording quality data. These devices could be worn for long periods of time (hours at a time), and are typically used in the context of child-centered research.

alecristia 326603780a correcting errors 1 rok pred
.datalad b391937f5a initial commit 1 rok pred
annotations 527a597b42 vtc annotations 1 rok pred
extra 7468c863e6 precisions.csv not a large file 1 rok pred
metadata 527a597b42 vtc annotations 1 rok pred
recordings cfd97a3d14 clean up 1 rok pred
scripts 6aeaafa0d0 compute precisions 1 rok pred
.gitattributes 7468c863e6 precisions.csv not a large file 1 rok pred
README.md 52a51722ed Mise à jour de 'README.md' 1 rok pred
datacite.yml 326603780a correcting errors 1 rok pred

README.md

Information on tests done comparing different recording devices

Information about the devices tested here.

Procedure

  • We created a "source" file with 5 mins from a CHILDES corpus (Thomas) and 5 mins from Seedlings -- this is the file called chain_inout.wav.
  • We played it from a mac computer with all of the devices in an array in front of the computer, in a sound-proof booth.
  • We tested 4 different device types + LENA+olympus (whose results I haven't calculated yet).
  • We then analyzed all of the sound files with our VTC and calculated % frames correctly classified (as FEM MAL CHI & silence -- there was no other child in these audios, an issue to improve in the next test!)

Conclusions

Based on this, our current recommendation is this particular USB, which is the cheapest yet leads to a performance above LENA's: MP power @ Clé USB Numérique Enregistreur Flash Memory Stick Voice Dictaphone Recorder 8Go 8Gb 8 Go

If you want to do even better, you can get this one: Enregistreur Vocal 16 Go, Enregistreur Micro Discret Numérique Activation Vocale, Dictaphone Portable Rechargeable par USB MP3 Player pour Cours, Réunions et Entretiens