Is there a way to record a conversation between 2 people and analyze the percentage of time each has spent talking?

Joe Videtto May 22, 2012
Pinterest Stumbleupon Whatsapp

Background:
I have a close friend, who I believe I may have insulted the other day by stating “We have a great relationship, because your capacity to talk about interesting things for an extended period of time matches with my endurance to listen for that amount of time…however, your capacity to listen is much shorter than your endurance for speaking”.

Ok…maybe not a feel-good comment to make to a close friend.

But – he said, “you just don’t realize how much time you spend talking”

Hence the question in the post:
Is there a way to use the computer to record a dialog between two people and later analyze it to determine the percentage of time each has spent talking?

Thanks in advance,
Joe

Ads by Google

  1. Kyem Ghosh
    May 30, 2012 at 7:35 am

    I got your logic and is really good but there is no way to differenciate two sounds from a single file untill there is a pre recognized data stored in the computer which stores the frequency and pitch. It matches the frequency (waveform) of the input voice with the one stored inside. But I have not known any software for this... High profile Voice recognisation device security softwares use this phenomenon. But there is no way to differentiate two unknown soundforms because when you record a conversation between two persons, it gets recorded to a single file and produce a single wave form. But there you'll get the veriation. The person with high frequency will have large bulging waveform (high amplitude) and the one having low frequency will have a small bulging waveform (low amplitude). But sometimes the person having low frequency may shout leading to higher frequency thus the waveform may get high bulging (high amplitude) for it. Or vice versa. Yes you can say that there should be a pitch difference between two people, but that too is stored in the wave design. Computer software has no ability to identify unknown pitches with its binary configuration. I'm studying on sound since last 2 years and to the best of my knowledge, I have created a document on the formation of soundwave of conversation of two people when recorded. Hope you'll understand it well. Here is the link: http://www.4shared.com/office/7dai_bmk/Sound_variation.html

    But you can do it manually with audacity... You can cut the voice of one person, copy and merge in a new file and hence you get the approximate time of the conversation. It obviously is time consuming but it will do..

  2. Oron
    May 23, 2012 at 11:39 am

    Joe,
    First of all, it's technically feasible, but I'm not aware of any "off the shelf" software package to do this. But if you want to do this with some manual input, check out academic speech analysis software such as the packages listed at http://www.surrey.ac.uk/sociology/research/researchcentres/caqdas/support/choosing/index.htm

    Secondly, you'd be missing the point if you went on to record your conversations with your friend, analyse them and prove him wrong (or even prove yourself wrong, G-d forbid!)... Surely the point is that you should find a way to get over the hurdle with your friend and not bring up the subject again!
    Good luck!

    • Joe Videtto
      May 25, 2012 at 1:33 am

      Actually, a little further self-reflection convinces me that he's right - I definitely have a greater capacity to talk continuously than I realized. I guess I just have to learn to say more with less words.

      And on your second point - your advice is well taken. I'm not really looking for a way to prove my point - because that would surely be missing the entire point.

      But what you may not have realized on this question is my desire to find an easy way to determine if my work with autistic children is resulting in an increase in the amount of time they spend communicating meaningfully, which is my goal. While the meaningfully part is pretty hard - I would think, as you said, the technology is there. I look forward to exploring your suggestion. Thanks.