I am planning to conduct a focus group study for a preliminary data collection exercise. I am planning to record (audio) the session as well for later transcribing. I have few questions;
1. How do you identify individual person’s discussion in the recording (audio)? In a focus group study, it is potential that more than one person speaks at a time. So how do you identify individuals? Are there any software or tools which identify the individual voices?
2. How do you minimise the outside noise (noises other than the participants discussions) during the focus group? I am more interested in software or tools in this question again.
There are audio editing software, you can identity individual if having sample of their voices.
2. You can remove backgroun noise but you should have 5 sec or more sample of background noise and get it filtered using software's and FFT
Thank you for your reply Akhilesh. Could you please name any software that can identify the individual voices?
The easiest way to achieve 1 is to use individual micrphone for each speaker, if you seek software solutions, the best would be using Speaker Diarization Software, such as VoxSort Diarization or CLEAVER (Oxford Wave Research)
For 2, the previous answer is correct, you can use software such Adobe Autidion o iZotope to process the audio
I agree with Andres. As well as using an individual microphone for each participant, record them on an individual track if possible. You can then use the level meters on each channel to identify the talker.
Another option for identifying talkers is to use a stereo microphone and listen on headphones. If the right microphone setup is used, such as an ORTF pair, it will sound as though you are in the room with the participants when listening on headphones and you can discern the talker by direction as well as the sound of their voice.
Whichever method you use, it is important to get the participants to identify themselves verbally at the beginning of the recording.
On reducing ambient noise in the recording, it helps to record in a quiet environment as it is very difficult to remove unwanted noise after it has been recorded. It also helps to use a smallish room that does not have too much reverberation (most radio studios fit this description). The third thing that will help is to use directional microphones (cardioid or hypercardioid) as these pick up less ambient noise than non-directional (omnidirectional) microphones. The fourth thing that will help is to keep the microphones close to the mouth of the talker. Headset mics are the best for this, although they can make some people nervous. Lapel mics are also good, and they are less obtrusive than headset mics. You don't need expensive mics for this - you can get quite usable lapel mics for about $20 each.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages
- Python