This page provides suggestions for researchers who wish to create audio files for analysis with the CLAN editor and other TalkBank programs.
For detailed phonological work, it is best to use 16-bit WAV format in stereo at either 22,050 or 44,100Hz. An even better format is 24-bit at 48K. However, for many purposes, recording in compressed formats such as MP3, AAC, or OGG is quite acceptable, as long as you use the high-quality versions of these formats.
Recording: There are five possibilities:
Decreasing Noise: Try to avoid recording near traffic or other external noises.
Rooms with carpets have fewer echoes and floor noises than rooms with hard floors.
Ongoing noise from televisions, radios, stereos, washing machines, computer fans, and other appliances can ruin a recording.
If noise is coming through an open window, consider closing it. Try to minimize spurious background noises or noises caused
from bumping the microphone onto hard surfaces or jostling the microphone cord. Try to keep the microphone out of clear view
so that the child will not try to talk directly into it.
Some researchers prefer to do their work in sound-proof rooms in a laboratory. Although it may be difficult for infants and mothers to adapt to the needs of getting to a laboratory, the results can be better acoustically. Although children produce less vocalization in such an environment, the nature of their vocalizations is similar, as has been demonstrated in Ledeweg et al. (1994).
You can also use the tape recorder as a notebook. You can record commentary directly onto the tape when the child is not talking and you are in a different room. You can begin each casette with a statement of the date, the year, the name of the child, the nature of the setting, and so on. This practice is very helpful in identifying tapes that have been otherwise mislabeled.