Voice identification plays a very important part in the forensic world. I practice voice identification regularly as an audio forensic expert and I believe it is both a viable and crucial science.
I chose this topic for this podcast because I just released my new book “That’s Not My Voice!: A Practical Understanding of the Art and Science of Modern Voice Identification.” In my book, I cover the basics of human hearing and my approach to voice identification.
How to Perform a Voice Identification
The process of voice identification is done through multiple steps.
The first step is critical listening. When you first receive a recording from a client, it’s important to listen through the recording in its entirety and take detailed notes on what you hear.
In order to continue with the voice identification process, an exemplar is needed as a comparison to the recording in question. Exemplars are supervised recordings of predetermined spoken word samples strictly made for voice identification comparison.
Electronic measurement has become possible with audio software that includes spectrum graphs. This allows you to gather notes on specific information on the frequency ranges of the two recordings.
Visual inspection of the waveform should be done next. During this step, you will analyze and compare the waveform of each recording. Words will have different lengths and varying levels for different syllables in the words.
Emergency voice identifications are cases where creating an exemplar of the voice in question is not an option. This is typically when a threat has been called into a company and they believe they know who called in the threat. In these cases, I must use any possible recording of the accused person such as a recorded voice mail or outgoing voice message. These are difficult cases but it is very important that the company stay in control of the situation before approaching the suspected person.