M12.3 Lesson 3 | HI-LING working page

Lesson 3: Voice Parade

Definition: A voice parade is a lineup of voice samples, from which an ‘earwitness’ identifies the possible perpetrator of a crime.

Key Concepts

Voice Parade
Voice similarity
Sequential procedure

UNIT 1: What is a Voice Parade?

A voice parade can be used as "earwitness" evidence in a criminal case if there is no eyewitness - the perpetrator was not seen and possibly could only be identified by his voice. This method is needed when a voice was heard at a crime scene but could not be recorded. Crimes like these involve a criminal wearing a mask, an attack from behind, or a phone call where the victim is threatened or blackmailed. In the process of investigation, the witness is asked to identify the voice of a suspect from a lineup. In the lineup, a sample of the voice of the suspect and the voice samples of people chosen by police are presented to the witness as a voice parade.

In the United Kingdom, voice parades are based on guidelines published in 2003. These guidelines were developed by adapting the police procedure for visual identity. Although voice identification is error-prone, and significantly less accurate than face identification, the evidence obtained can be crucial in influencing the outcome of a case.

Voice parades ideally consist of nine voice samples, 8 from people chosen by police and one from the suspect. The voice samples from the people chosen by police are known as foil samples. They have to go to the police station and participate in a recorded interview. The recorded interview is cut down to a collage of short utterances of spontaneous speech with a duration of 45-60 seconds and will then be presented as one of the 9 voice samples to the witness. Later on, the witness hears all the foil samples and the suspect’s voice sample over a video or PowerPoint presentation and then tries to point out the perpetrator’s voice after hearing them all. Due to the time-consuming nature of this procedure and limited resources, police often use fewer participants and not an interview-styled collage recording. They let the participants of the voice parade repeat the words the witness heard during the crime. This way less time passes between the crime and the witness having to identify it.

Activity 1: Comparing

Voice Samples

Try to match the following 4 recordings into pairs. Talk with your partner, about which two are from the same person and why you think that.

Set 1

Set 2

Voice 1.0

00:00 / 00:10

Voice 2.0

00:00 / 00:05

Voice 1.1

00:00 / 00:12

Voice 2.1

00:00 / 00:06

UNIT 2: Voice Similarity

The people giving foil samples should have a similar voice to the suspect so voice similarity is guaranteed. If not, the witness could focus on a singular characteristic he noticed from the perpetrator’s voice. For example, if the perpetrator had an Irish accent the police suspect would also have an Irish accent. If that suspect is now lined up with foil samples of British accents, the voice parade loses credibility because distinguishing the dominant characteristic the witness noticed at the crime from the other voices is very easy. A much more effective voice parade would be if all eight foil samples presented had an Irish accent. This way, police have a higher certainty that the suspect is the perpetrator and not one of many speakers who have an Irish accent. This way the false identification through a dominant voice characteristic can be eliminated.

Voice similarity is crucial and the most challenging aspect of constructing a voice parade. It involves on the one hand linguistic factors related to language, dialect, and accent. On the other hand, personal factors that are related to an individual's anatomy and physiology when producing speech. These phonetic properties include in particular pitch, speaking rate, resonance features, and voice quality. Whereas pitch and several voice quality features linked to the larynx (cylindrical grouping of cartilages, muscles, and soft tissue that contains the vocal cords) and pharynx (muscles that connect the nose and mouth to the larynx and esophagus) are most important for voice similarities.

The voice parade gets more difficult for a witness the more similar the recordings are and the more likely a wrong voice sample can be chosen by the witness. During the investigation, a person who gives a foil sample would not be charged with the crime if the witness identifies them as the perpetrator but it would help the suspect’s exoneration. As a consequence, the prosecution could not use the voice parade as evidence in court.

Voice parades are still rather new to law enforcement and how their methodologies can be improved is still being researched, for example at the phonetics laboratory of Cambridge University.

In this scenario, an employee of a cheese factory was staying late due to having to work overtime. Suddenly she sees a shadow under the office door so she hides behind her desk. The door opens and quick steps are heard walking to the other side of the room where the confidential files are stored in the cheese vault. A phone rings and the perpetrator answers it. “Yes I’m in, nobody is here but I don’t have much time and yes I won’t forget to buy groceries tomorrow”. The perpetrator cracks the vault and leaves with the confidential files. The employee calls the police. Scared to death she recalls what had happened and what she heard.

Detective X is on the case but they know they must act quickly, as a suspect has been caught near the crime scene. Time is of the essence because a witness’s memory can be affected and the sooner the better that a voice parade takes place.

You will now hear the recording as it would have been during the theft from the point of view of the employee but only once. In the next step, you will have to identify the perpetrator from a voice parade of four voice samples. Good luck!

Activity 2: Case of

the Stolen Cheese

Sample 1

00:00 / 00:07

Sample 2

00:00 / 00:13

Sample 3

00:00 / 00:14

Sample 4

00:00 / 00:12

Original

00:00 / 00:07

Final thought for this lesson

Here are the key takeaways from today's lesson:

With a voice parade an earwitness can help to identify a criminal but an exact procedure is necessary to reduce error margins.
Voice similarity is a key concept to regard in a correct voice parade

Do you think you would be a good earwitness and able to identify a criminal in a voice parade?

Sources:

Cambridge Phonetics Laboratory (October, 2023). Retrieved from from: https://www.phonetics.mmll.cam.ac.uk/ivip/ivip-project
McDougall, K. (2013). https://www.thebritishacademy.ac.uk/documents/805/BAR21-06-McDougall.pdf
Smith, H. (2020).http://irep.ntu.ac.uk/id/eprint/37747/1/14891_Smith.pdf