I used an expert voice analyst in one of my cases last year. I’ve been thinking about writing about the experience for a while, as it left me with grave doubts about the supposed science of voice elimination. For some reason, I started thinking about it again over the holidays and finally decided to write something. It seemed like a good way to kick off the new year and get back to blawging.
The case involved an extensive, multi-jurisdictional wiretapping investigation. My client was accused of conspiring to purchase drugs and using a wire communication to facilitate a felony drug crime. He claimed it wasn’t him on the phone. We wanted to use an expert to eliminate my client as a potential match for the voice on the recording.
I inherited the case from another defense attorney who had already gotten the court to appoint a forensic voice expert. I contacted that expert, but it turned out he was unable to handle the case. When I asked him for recommendations, he gave me a few names. I did a fair amount of research on my own and ended up settling on one of those experts.
The man I hired had an incredible resume, and I got a good feeling speaking with him on the phone. He was extremely professional and seemed to know exactly what he was doing. I could see why he came so highly recommended.
I sent the expert the recordings he was supposed to analyze as well as the notes and synopsis of the agent who was listening in on the calls. I thought that would help him navigate through the conversation. I also described to him on the phone and via email what my client was alleged to have said in the conversation in question.
When he sent me the exemplars he wanted my client to read, he only indicated that I should contact his office to set up a time for my client to perform them. They were very simple exemplars, mostly single words like “yeah” or “alright.” It wasn’t a verbatim transcript of what my client allegedly said, but it was close. The exemplars included words that both speakers used as well as one phrase only the speaker not alleged to be my client used. I believed the expert was using that to compare my client’s voice to a known negative.
My client performed the exemplars as planned, and I received the expert’s report. It was inconclusive but suggested my client was somehow altering his voice to avoid a match.
The expert’s “aural evaluation” noted inconsistency relative to pitch and syllable and consonant production. Basically, he thought the frequency of my client’s voice varied too much. To him, that suggested my client was trying to disguise his voice.
The “spectrographic evaluation,” which the expert claimed measured variables people could not manipulate, showed consistency between my client and the unknown speaker. That portion of the report seemed very scientific, with exhibits, charts, and complex computer-generated representations. The visual comparisons looked very similar.
Basically, the portion of the test that most people would trust as being scientific and therefore infallible showed a match. That was the portion the expert claimed my client could not affect by altering his voice. The portion of the test he could affect showed inconsistent speech, indicating deception.
Here’s the problem: I believed my client when he insisted it wasn’t him. I thought the voice on the recording sounded nothing like my client, and I heard him perform the exemplars and thought he read them exactly how he normally spoke. He has a very distinctive voice and speaking style, but it didn’t seem like he was disguising anything.
I spoke with the expert and tried to figure out how certain he was about the spectrographic analysis. He was pretty sure. I explained my concerns and mentioned my client’s insistence that it wasn’t him.
At some point in the conversation, we discussed the exemplars. I asked whether the results might be different if he used a verbatim transcript of the conversation for comparison. I asked him if the known negative he included in the exemplars also showed a match on the spectrograph.
The mood of the expert changed completely when I asked that second question. He told me there was no known negative. My mood changed too. I think we both realized at the same time what had happened: he did the analysis on the wrong speaker.
It took us a little while to figure out how he ended up doing the analysis on the other voice. It turned out that a portion on the reviewer’s notes from the prosecutor’s disclosure inverted the known speaker’s name and the unidentified voice alleged to be my client. That erroneous portion was what the expert relied on in determining which voice he was supposed to identify. The mix-up wasn’t clear from the exemplars because each of the speakers said such similar things in the short conversation.
I felt like a moron. The expert was not happy. More than anything, he seemed extremely concerned with the whole situation. His report may have been inconclusive, but if he was the prosecution’s expert witness at trial, his testimony probably would’ve been more than enough to convince a jury that the speaker was my client and he just disguised his voice to avoid a match. I don’t think that fact was lost on him.
The expert agreed to redo the test, and we started the whole process over again. The second time, however, the expert’s finding was a probable elimination. According to his second report, the unknown speaker in all likelihood was not my client.
The experience left me very skeptical about voice identification. It made me wonder if other supposedly scientific technologies that make their way into the courtroom are as problematic. I wasn’t shocked when I read this report from the NAS on issues with forensic analysts. If my expert, a world-renowned expert in his field, got a false match, why should I trust a government employee with the minimum training and college science credits necessary for the job?
As a lawyer, I’m used to dealing with gray areas. After a long day of answering every question with “maybe” or “it depends,” it would be nice to know I have a scientific expert who can tell me “yes” or “no.” Other lawyers probably feel the same way. Most lawyers, prosecutors and defense attorneys alike, are comforted by the idea of a detached scientific expert making their case with an unambiguous answer. We should be more skeptical.
In the end, the case against my client got dismissed prior to trial. The case was weak to begin with, and the expert’s second report was the final nail in the coffin. I was a little surprised the prosecutor ultimately trusted the expert’s report, but maybe she just really wanted a definite answer. She’s one of the fairest and most experienced prosecutors I work with regularly, so it may be that the report, while not perfect, was just enough to give her a reasonable doubt. She could have had a field day with him at trial.
Maybe the thrill of science-like evidence hadn’t faded for her. Maybe she listened to the recording again and compared it to my client’s manner of speech. It was so clearly not him that I still think the case should have been dead in the water a lot earlier. Instead, we used an expert whose expert technology almost got a match for the wrong guy. Are we so enamored with science in the courtroom that we abandon our own ears and common sense in favor marginally-accurate technology? No need to respond, I know the answer.
My client was out of custody and suffered no negative consequences as a result of the mix-up. He was pretty stunned by the whole expert debacle and thinking about my conversations with him ultimately convinced me to post this. He was happy to be done with the case, and so was I.
Although the case is over now, I doubt I’ll quit thinking about the experience. I know I’ll never look at any expert the same way.
Tags: aural evaluation, dismissed, elimination, expert, spectrographic evaluation, use of a wire communication, voice analysis
This is a fantastic post. Thanks for sharing. So much of the science is not scientific at all. Even when the science *is* there, near-perfect communication is required to make sure everyone is on the same page.