Voice Analysis Technology

[Fred F.]

This is an article from a Las Vegas business paper regarding a "new" voice analysis technology being developed by Mr. Harry Rosen of LEA Detection Systems. If CVSA scares you, this will scare you more. The LASD is "testing" this technology with homicide suspects to save man hours.

here is a quick peek

Patented in 1999, LEADS was born from decades of research by Asian psychologists       who were looking to identify slight discrepancies in speech patterns when patients were         deliberately trying to deceive during an interrogation. As a result, the software has more          than 18,000 recorded voices installed into it and uses precise algorithms to detect even            the slightest variation in speech that highlight whether a patient or inmate is attempting           to deceive with his or her answers.

Anyone heard of this before ?


http://www.lvbusinesspress.com/articles/2003/06/27/business_profile/busproleads.txt



Fred F.

[Twoblock]

Fred F

I wonder if Harry Rosen would care to list his mining and oil exploits. This sounds like it is another one of his promotion schemes. After all it sounds like he used someone else's product and supposedly "improved" it. This scenareo sounds familiar. I am, also, familiar with mining and oil promotions.

Anyone with a trained voice, such as a singer, could knock the hell out of any voice analysis instrument. As well as being a miner, I am also a singer. I can hit an A4/40 C on demand without a pitch pipe.

I wonder if LEA Detection would put up a 10 to 5 deal like the polygrapher in Texas?

[orolan]

There was a write-up about this for a venture-capital symposium in New Mexico. Interestingly enough, DOE's Sandia Laboratory is a primary funding source.

LEA Detection System, Inc. brings a unique, patented technology to the law enforcement and security markets.
A top priority of these agencies is to have a reliable means of determining deception in human interactions. A
computer program called LEADS analyzes elements of actual or recorded speech that reflect neurological intent
to deceive. LEADS' determination of deception is independent of language spoken and is beyond the speaker's
control. LEADS will be marketed through companies that service law enforcement agencies and security companies.
A pilot program with the Los Angeles County Sheriff's Department is underway. Equity capital of $1
million is needed to begin business operations, followed by additional funding to underwrite operations during
the ensuing eighteen months.

http://www.techventures.org/2003Presenterbrochure.PDF

[suethem]

In a time when we need 'heads on sticks' and more 'boots' on the ground in both LE and Mil- this is what we get.  What a waste of funds.  

Maybe I should get in on the con.  My new 'investigative tool' could be  called "the suspect wheel."  Spin and look at the 'indentifier window' and find your man.  

Don't worry about knowing your neigborhoods or making contacts in the community- 'the suspect wheel' takes away all those lingering questions and gives you the easy answer that seemed so difficult before.  Racial profiling built right in !!!!  Its so easy and cheap!!

[Fred F.]

Fred F

I wonder if Harry Rosen would care to list his mining and oil exploits. This sounds like it is another one of his promotion schemes. After all it sounds like he used someone else's product and supposedly "improved" it. This scenareo sounds familiar. I am, also, familiar with mining and oil promotions.

Anyone with a trained voice, such as a singer, could knock the hell out of any voice analysis instrument. As well as being a miner, I am also a singer. I can hit an A4/40 C on demand without a pitch pipe.

I wonder if LEA Detection would put up a 10 to 5 deal like the polygrapher in Texas?

TwoBlock,

Rosen says

We're trying not to oversell it -- but it works," says the World War II veteran

He did manage to get the LA County Sheriff's Dept. to use this "technology" and have "results" of weeding out "innocent" people. What happened to investigations done the old fashioned way? Too many criiminals getting away?

I also would like to see how his mining and oil exploits went as well. He obviously didn't do too well.

Can you explain the voice pitch A4/40 C? Rosen claims to have 18,000 voice "patterns"but you would think that there are far more than that.


Fred F.

[Twoblock]

Fred F

A440 means that the wave length contains 440 beats a second at sea level. In the mountains of Colorado it would be something like 442. An A440 tuning fork at sea level would be off at 6000 ft. because of the different atmospheric pressure. Thefore my voice C would be off. Is Rosen going to keep his subjects at sea level while testing?

As to the 18,000 voice patterns, I would think there would be many more. Chet Atkins, my guitar idol and who I patterned my guitar playing after, knew over 2500 different chord variations. Each has many voice variations.

I just don't buy voice analysis of any kind.

[Marty]

Twoblock,

Very interesting! I had never considered the impact of altitude on pitch. I assume common practice would be to adjust string instruments (which wouldn't be changed by altitude) to match the pitch shift of wind instruments since it would be presumably harder (or impossible) to tune them for the shift in the speed of sound.

-Marty

[Fred F.]

TwoBlock/Marty

It seems that this "technology has more "holes" in it then Mr. Rosen wants you to know.

Another thing that puzzles me since you say voice is affected by atmospheric and other conditions, How can he claim to have success analyzing recorded voices?  If Chet Atkins has 2500 chord variations, then it would be safe to assume that he has more voice pitch variations?

TwoBlock, Do you have an educated guess to the number of pitch variations in the average person?

Fred F.

 

[orolan]

Twoblock,
My daughter has a classmate in her Applied Voice class who has Spasmodic Dysphonia, the Adductor variant. Her voice is all over the scale, varying in pitch, loudness, hoarseness, etc. And the more "stressed out" she gets, the worse it gets. This girl went to 8 different doctors over a 10-year period before her condition was properly diagnosed. My daughter ended up writing a paper about the girl for her Psych class, which is how I found out about it.
What do you know about this, and doesn't it seem that a person with this affliction is guaranteed to fail any type of voice-stress analysis?

[Twoblock]

Fred F.

I would guess the avg. person speaks in the 6 to 8 whole note range. Some people speak in monotone. Singing range much different. For instance -  Johnny Cash has maybe 1 1/2 octave singing voice and Opera singers like Marilyn Horn and Maria Calas had maybe 3 1/2 octaves. (Some singers sing in monotone).

Marty

Wind instruments are easily tuned. Years ago I used to play trombone and some days, probably when tired, I would play flat. For that particular concert I would have to tune "up" slightly. The tuning mechanism is at the back of the isntrument.

BTW, for people not musically inclined, the A in front of the 440 means the key of A.

Bottom line. I would say voice analysis has, maybe, the validity of CQT polygraph. It MIGHT be able to gain a confession simply out of fear of the instrument.

[Twoblock]

Orlan

I know nothing about Spasmodic Dysphonia. However, in my years of performing, I have known a couple of would be performers like your daughter's friend. Both were good when singing in their living rooms but, put them on stage, in front of an audience (stage fright stress) and every note was extremely sharp. One was able to overcome her stage fright and went on to become a good singer. The other never could overcome her affliction and chose another profession. Back then, I don't think doctors had a name for it.

My opinion is: Yes, these people would be set for failure in voice analysis tests. Stress, fear, anger etc. usually makes people's voices get a little higher. These factors will affect the outcome of the polygraph or any other "truth finding" machines.

I would like to read your daughter's paper.

[Fred F.]

I would guess the avg. person speaks in the 6 to 8 whole note range. Some people speak in monotone. Singing range much different. For instance -  Johnny Cash has maybe 1 1/2 octave singing voice and Opera singers like Marilyn Horn and Maria Calas had maybe 3 1/2 octaves. (Some singers sing in monotone).

Twoblock,

I wonder how Mr. Rosen can claim to have such success with TAPED voices. Can a tape capture the true essence of ones voice? I am a bit skeptical of LEA Detections claim that they can get accurate "readings" from tapes. With the advent of CD's MP3's and digital recording which provide far more accurate recording than tape, Rosens claims seem a little far fetched.

In regards to monotones, are there measureable pitch variations?

Orolan,

I would concur with you on your daughters friend. Maybe you can contact LEA and ask them if  
Spasmodic Dysphonia, the Adductor variant will affect the "readings"


Fred F.

[Twoblock]

Fred F

Our so called "experts" in the FBI, CIA. etc. can't even accurately identify the voices on tapes attributed to Been Layen and Hishiney so how in hell does Rosen think he can tell if people are telling a lie from tapes. Sounds like another Wonder Woman lasso to me. Tapes capture so much background distortions.

Monotone means without change of pitch or key. one tone. People who talk this way would probably drive Rosen and his people crazy. Can you imagine what a single pitch sound wave staring them in the face would do to their egos? Would they acuse the subject of using countermeasures? Probably so.

[orolan]

I keep getting this picture in my mind of Rosen trying to examine the comedian Stephen Wright. Those of you who have seen him know what I'm talking about. For those of you who haven't, he tells all of his jokes in a very "deadpan" monotone, with no facial expressions whatsoever.

Twoblock, I'll see if my daughter still has her paper on her computer.

[guest from canada]

I have to laugh!  I just checked out the polygraphplace web page.  What a joke!  One interesting thing I found was a statement there that regards the voice analysis technology as not having scientific support.

"Instruments that claim to record voice stress are not polygraphs and have not been shown to have scientific support."

Hee, hee.  The anti poly people here have nothing to worry about.  Just sit back and watch the voice technology and the pro-poly sides start tearing each other apart as they both vie for dollars and a foothold in the tea leaf reading/astrology/truth detection market.

Since when did the poly GAIN scientific support???