17 yo Trayvon Martin Shot to Death by Neighborhood Watch Captain #13

DNA Solves
DNA Solves
DNA Solves
Status
Not open for further replies.
protestings, riots, etc. what ever the hell they are called.

So sorry I just happen to mess up.

Riots denote violence and destruction of property. I'm just not seeing that. And it's to many people's credit that they are holding their EXTREME frustration in check in the face of some of the most repugnantly racist comments they hear. This restraint is a HUGE part of what won over the hearts and minds of formerly bigoted "middle" Americans during the influential years of Dr. Martin Luther King.
 
protestings, riots, etc. what ever the hell they are called.

So sorry I just happen to mess up.

<modsnip> Here's a good way to tell the difference. Look at he pictures of the protests in Sanford this week. Then GOOGLE: "Chicago 1968", "Watts Riot", "Rodney King riot". Look at those pictures and compare to what you've seen this week.
 
BBM.
I don't think so. It can be twisted but it's not what they're saying. It's not a percentage of certainty. It's the percentage of attributes that they're measuring that match Zimmerman's.

All human voices probably share some percentage of attributes. Male voices share a higher percentage of attributes.
They said a 48 % match means that they're reasonable certain it's not the same person; saying they're 48 % certain would just mean that they don't know either way.

Strawberries and eggplants both have 92 % of water but we all can tell them apart with a 100 % certainty.

BBM
Why do male voices share a higher percentage of attributes?
 
lol, I get that. But these were also very clear recordings and only got 86% when this expert's threshold is 90%. This does not add up to me.

In technologies that are based on matching it is always a trade between sensitivity and specificity.

If you put the threshold low, you are going to be more sensitive and be able to match more matching samples. But you're also going to be less specific and get more false positives.

If you put the threshold high, you are going to miss some matches but you are going to get fewer false positives.

Then you have to balance the probabilities and decide what threshold gives you the best result.

Keeping in mind that if you are going to use this technology in court contexts false positives could get innocent people convicted.
 
So, m00c0w, are you saying there was a THIRD person at the murder scene who was doing all the blood-curdling screaming?

If the screamer was NOT Zimmerman who do YOU suppose it was?

imo

How accurate is this technology? Just because Zimmerman scored 48% does it necessarily mean he is excluded?
 
BBM
Why do male voices share a higher percentage of attributes?

That was badly worded of me.
I mean, male voices are generally more similar with other male voices than with female voices.
 
BBM.
I don't think so. It can be twisted but it's not what they're saying. It's not a percentage of certainty. It's the percentage of attributes that they're measuring that match Zimmerman's.

All human voices probably share some percentage of attributes. Male voices share a higher percentage of attributes.
They said a 48 % match means that they're reasonable certain it's not the same person; saying they're 48 % certain would just mean that they don't know either way.

Strawberries and eggplants both have 92 % of water but we all can tell them apart with a 100 % certainty.
I don't know. I really wish there was some documentation for this software somewhere. I'd actually like to get my hands on it myself. Unfortunately I don't have $5k to blow on it, sadly.
 
Riots denote violence and destruction of property. I'm just not seeing that. And it's to many people's credit that they are holding their EXTREME frustration in check in the face of some of the most repugnantly racist comments they hear. This restraint is a HUGE part of what won over the hearts and minds of formerly bigoted "middle" Americans during the influential years of Dr. Martin Luther King.

Only the one riot so far. That was in Miami where they looted and ransacked the walgreens.
 
In technologies that are based on matching it is always a trade between sensitivity and specificity.

If you put the threshold low, you are going to be more sensitive and be able to match more matching samples. But you're also going to be less specific and get more false positives.

If you put the threshold high, you are going to miss some matches but you are going to get fewer false positives.

Then you have to balance the probabilities and decide what threshold gives you the best result.

Keeping in mind that if you are going to use this technology in court contexts false positives could get innocent people convicted.
I mean that the expert considers 90% a match, but when the person demonstrating compared two recordings of the same person (Nixon), it only displayed an 86% match. I guess I'm not understanding how he expects to get over 90% when two very similar recordings were compared and only came out as 86%.
 
How accurate is this technology? Just because Zimmerman scored 48% does it necessarily mean he is excluded?

I think it's been around a long, long time. I think the shhhhhhh, (whispering) that the C_I_A may use it to identify targets. I really do not know but I do think it has been around for a long time and I'm sure the government has probably used it more than once or twice. jmo
 
Peeps!

Please do NOT get into personal back-and-forth with one another.

In my experience that is a quick way for any thread to go right off the rails.

If you find yourself unable to address a fellow poster without snark or aggression, then either do your best to de-personalize your response, or just take a moment to calm down--BEFORE you hit Send.

Also, remember to use your Alert button rather than responding to a post that violates TOS.

This post lands at random, and is only intended to help you continue what has been a fairly civil discussion tonight.


:tyou:
 
The person demonstrating the product was getting an 86% result with known recordings of the same person...

So either he rounded up or he was holding himself to an even higher standard. Neither possibility invalidates his conclusion that 48% is not a match.
 
So either he rounded up or he was holding himself to an even higher standard. Neither possibility invalidates his conclusion that 48% is not a match.

Well if Trayvon's voice scored higher at least we would have something. As it is, we don't even know if Trayvon was going to be a match with this method.
 
I mean that the expert considers 90% a match, but when the person demonstrating compared two recordings of the same person (Nixon), it only displayed an 86% match. I guess I'm not understanding how he expects to get over 90% when two very similar recordings were compared and only came out as 86%.

Nixon, in like Richard????
 
If your saying that the Orlando Sentinels voice analysis article is not the "smoking gun" that leads to GZ's arrest and conviction, I agree.

I agree with you there. But my point was that the one sentence quote we have heard re the 48% is not really a fair indicator of what a jury will be told in a courtroom.
 
Status
Not open for further replies.

Members online

Online statistics

Members online
55
Guests online
2,239
Total visitors
2,294

Forum statistics

Threads
601,855
Messages
18,130,768
Members
231,162
Latest member
Kaffro
Back
Top