Can you hear me?

No Gravatar

Imagine this scenario.  It’s almost Christmas.  And, you’re in a Hilton hotel bar- teeming with people celebrating, virtually wall to wall.  And, you hear a voice.  And, recognize it as that of a friend from decades past.  “Jeffrey Fried, where are you?”, I call out.  Seconds later… “Roy?”.   I hadn’t seen Jeff in 21 years.  But, I recognized his voice.  (Which restarted our friendship for another decade, until I got married and moved away.  Jeffrey Fried- where are you?)

That scenario describes one that scientists term the “cocktail party effect”  (the ability to pick out a voice in a crowd).     And, Drs. Nima Mesgarani and Edward Chang (UC San Francisco) advanced the state of the art using three epileptic volunteers, as they reported in Nature.  These three subjects had electrode arrays already implanted in their brain to help control their seizures.  (We’ve discussed this concept before… Deep Brain Stimulation, where patients agree to have electrodes implanted in various parts of their brain to stimulate different reponses.)

Drs. Mesgarani and Chang were able to monitor the subjects to determine which neurons were active in response to specific sound frequencies.  From that data, they were able to produce a spectrogram- a print-out of the sound properties and the time frame of occurrence.  (These graphs are sometimes called voicegrams or voiceprints.)

Locus for voice recognition in brain and voiceprints

Once these spectrograms were developed, the researchers went about examining the cocktail party effect.  The subjects were asked to listen to voice recordings from a man and woman.  Their task was to discern when a certain key word was spoken by either of them (they were not told which person would employ the word)- and then focus only on that voice from then forward.  And, the researchers found by monitoring the subject’s brain activity, they could discern which voice the subjects were “hearing”.    This was true whether the subjects were paying attention  to the correct speaker or not.

The next goal for this research  is to refine the algorithms so that voice recognition systems can decipher specific speech in situations where there are multiple speakers.  Given that solution, we can have computers find that voice in a crowd- or transcribe conversations or speeches with significant background noise.

Share this:
Share this page via Email Share this page via Stumble Upon Share this page via Digg this Share this page via Facebook Share this page via Twitter
Share

32 thoughts on “Can you hear me?”

  1. Amazing. I have never really thought about how we might recognize the voices we know even in a crowd. Science is exploring some interesting aspects of how our brain works. However Roy, I’m with you on the idea of having electrodes implanted in my brain…creepy even if it does help some.
    Bonnie recently posted..An Important Truth About Mindset

    1. That technology is very different. We have been able- with varying degrees of success- to dictate into computers (and since smart phones have computer chips, for the purposes of this discussion, Lynn, they are computers as well) after we have trained the application to recognize of speech patterns.
      This application detects sounds among a cacophony- which means, for example, if the computer HAS been trained to identify your voice, it could transcribe your remarks among all the other conversations. Or, if all the speakers’ patterns have been identified, a complete transcript of a meeting could be automatically produced!

      Roy

  2. I’m with you, there shall be no electrodes in my brain while I still have breath in my body. But the innovations are amazing, the growing understanding of our minds is amazing…10 years from now I think we’ll look back to now and be amazed at how primitive our understanding was. Kind of like how we look at the early 1900s. Love the information!

  3. Interesting, Roy. I hope you find your friend again. For my mom’s 70th birthday we had a surprise party. My sister from WA flew in and infiltrated the party mid-stream. She walked up behind my mom, who is quite sharp by the way, and said “Happy Birthday.” Mom didn’t recognize her voice at first. I guess sometimes the opposite is true – you know a voice very well but know it doesn’t belong where you are and your brain doesn’t register.
    Bonnie Anderson recently posted..Please Read This Yesterday

    1. Oh, Bonnie- that is not generally the case. It could be that the noise in the party was at the same frequency as your sister’s voice, which might have masked it. Otherwise, I would aim for a quick hearing check… Really.

    1. Alessa:
      The problem with your smart phone is that it may be “smart” when compared to another phone, but it’s IQ is limited by the strength (processing power) of the chip that drives the unit. This is akin to the issue of the first speech recognition systems that were relying on 90286 chips. The science was there- but the processing power was not, so more errors occurred. As the “brains” of the units (the processing power) increased, the recognition capabilities increased dramatically.

      Roy

  4. Dare I say it sounds very similar to how content is found online using keywords?

    In order to break up the noise a user on the internet types in search term or keyword in to a search engine.
    That search engine filters out the surrounding noise and presents back to the user a list of potential conversations or messages to pay closer attention to.
    James Debono recently posted..Marketing with Social Media – Grow A Loyal Community by Increasing Your Worth!

    1. James:
      One could consider that akin- but I believe the search engines are different in that they are already programmed to seek out the words chosen for keywords and/or to read the titles of a page.
      In this case, it’s closer to a needle in a haystack in that we don’t know for what we are looking, but recognize familiarity once we hear it. Using a search engine similarity, it would be akin to finding the name of the author for each page, since their “voice” would be understood from the vernacular, idioms, and structure.

      Roy

  5. Pingback: Homepage
  6. Pingback: site
  7. Pingback: v2 electronic cigarette coupon
  8. Pingback: הובלות
  9. Very nice post. I just stumbled upon your weblog and wanted to
    say that I have truly enjoyed browsing your blog posts.
    In any case I will be subscribing to your feed and I hope you write again very
    soon!
    Jacob recently posted..Jacob

  10. Do you mind if I quote a couple of your posts as
    long as I provide credit and sources back to your webpage?
    My blog is in the exact same area of interest as yours
    and my users would definitely benefit from a lot of the information you provide here.
    Please let me know if this ok with you. Many thanks!
    best mens cologne ever recently posted..best mens cologne ever

Comments are closed.