Snap Shot
- The question: How reliable is a witness’s identification of a person by voice, and what conditions and procedures make it more or less trustworthy?
- What the research shows: Familiar voices are often recognized well, though not without error. Identification of an unfamiliar voice heard once is generally less accurate than identification of an unfamiliar face, and it is affected by exposure, delay, disguise and whispering, changes in tone, accent and language, and telephone transmission. Confidence is an inconsistent guide for voices, and lineup procedures, especially pre-lineup instructions, measurably change error rates.
- What it does not show: The research does not show that voice identification is impossible or that any particular identification is right or wrong. Most studies use brief recorded voices, student or online listeners, and no real consequences, and the field has far fewer studies than visual identification.
- The law: A lay witness may identify a voice under Fed. R. Evid. 901(b)(5), Mil. R. Evid. 901(b)(5), and Ariz. R. Evid. 901(b)(5) on a minimal showing of familiarity, leaving reliability to the factfinder; suggestive voice procedures are reviewed under the due process framework (Goudeau). Expert voice comparison is tested under Rule 702; in United States v. Angleton, defense-offered aural-spectrographic testimony was excluded.
- Issues for each side: Prosecutors can show favorable listening conditions and a well-run voice lineup, without treating an earwitness’s certainty as proof; defense counsel can show the specific conditions that reduce accuracy, without implying that voices can never be recognized.
An earwitness is a person who heard, rather than saw, someone involved in an event and later tries to identify that person by voice. Voice identification arises in threatening and ransom calls, robberies by masked offenders, assaults in darkness, and recorded calls in which a speaker’s identity is disputed. A witness may say a voice belonged to someone already known, or may be asked to pick a stranger’s voice from a voice lineup, often called a voice parade.
Voice identification shares with visual identification the distinction between estimator variables (conditions of the original hearing) and system variables (how the identification is obtained), and much of the legal framework. But voices are hard to describe, one speaker’s voice varies greatly, and the research base is far smaller. This paper covers familiar and unfamiliar voices, conditions that affect accuracy, confidence, voice-lineup procedures, memory for what was said, forensic voice comparison, and the federal, military, and Arizona law.
Familiar and Unfamiliar Voices
The most important distinction is whether the listener already knew the voice. Neuropsychological, clinical, and experimental work indicates that recognizing familiar voices and discriminating unfamiliar ones rely on partly different processes; familiar-voice recognition is comparatively robust, while unfamiliar-voice memory is weak and varies widely across listeners (Stevenage, 2018). One reason is that a voice varies with emotion, speaking style, and context. Listeners who know a speaker treat that variation as one identity, whereas listeners who do not often mistake two samples of one voice for two people, or two similar voices for one (Lavan et al., 2019).
Familiarity is a matter of degree and does not guarantee accuracy. When listeners tried to name personal associates from recorded speech and to reject a stranger’s voice, accuracy depended on how well the speaker was known and whether the speaker whispered, and beliefs about voice-recognition ability exceeded performance in most conditions (Yarmey et al., 2001). A claim of familiarity invites questions about how often, how recently, and in what circumstances the witness had heard the speaker.
How Accurate Is Voice Identification?
Research on unfamiliar voices, beginning with McGehee’s (1937) experiments, has long reported modest accuracy and frequent false identifications (Clifford, 1980; Yarmey, 1995). The Lindbergh kidnapping trial is often mentioned as background: Charles Lindbergh heard a few words called out at night and, more than two years later and after Bruno Hauptmann’s arrest, identified Hauptmann’s voice. It illustrates brief exposure, long delay, and knowledge of the suspect; it does not measure error rates.
Recent studies reach similar conclusions. With nine-voice parades, participants picked the correct voice 39% of the time when it was present and correctly rejected the parade only 6% of the time when it was absent (Smith et al., 2020). In two experiments with about 540 participants, performance on target-absent parades was at chance levels (Pautz et al., 2023). Faces were recognized better than voices in a direct comparison, and seeing a speaker’s face interfered with later voice recognition, though not the reverse (Stevenage et al., 2011); for a witness who both saw and heard a speaker, the voice may be the weaker memory.
Accuracy is not uniformly poor. Under favorable conditions, such as matched studio-quality recordings, most listeners in one study identified a target a week later (McDougall et al., 2015), and longer exposure, distinctive voices, and familiarity all improve performance. The conditions of the case matter more than any general figure.
Conditions That Affect Accuracy
Exposure matters. In a mock-kidnapping study, listeners who heard the voice for 8 minutes identified it more accurately two days later than listeners who heard it for 30 seconds (Orchard & Yarmey, 1995). Reviews of the earlier research describe delay as a factor in unfamiliar-voice accuracy, although studies vary in how quickly memory declines (Yarmey, 1995).
Disguise and changes in delivery reduce accuracy. In the same study, whispering, the distinctiveness of the voice, and a change in tone between hearing and lineup all affected identification on suspect-present and suspect-absent lineups (Orchard & Yarmey, 1995). A voice heard in anger may be harder to recognize in calm speech (Saslove & Yarmey, 1980), and whispered speech reduced recognition even of known speakers (Yarmey et al., 2001).
Language and accent matter. Listeners identify speakers better in a language they understand (Goggin et al., 1991; Philippon et al., 2007). Within a language, English and Scottish listeners were each more accurate with voices in their own accent, in target-present and target-absent lineups (Stevenage et al., 2012). This “other-accent” effect parallels the cross-race effect for faces and bears on filler choice when witness and speaker come from different linguistic communities.
Transmission matters, especially when conditions do not match. Accuracy was 76% when exposure and parade both used studio-quality audio and 64% when both used telephone-quality audio, but only 32% when the voice was first heard by telephone and the parade used studio recordings; confidence did not reflect these differences (McDougall et al., 2015). Lineup recordings should match, as closely as possible, the way the voice was originally heard.
Confidence and Accuracy for Voices
For faces, high confidence on an initial, fair test is associated with high accuracy; for voices the relationship is weaker and less consistent. In a comparison of eyewitnesses and earwitnesses, earwitness identifications were less accurate and earwitnesses were more overconfident (Olsson et al., 1998). Later studies have reported small but significant confidence–accuracy correlations (Orchard & Yarmey, 1995), confidence that was diagnostic for some levels of familiarity (Yarmey et al., 2001), and confidence that did not track accuracy (Smith et al., 2020). Confidence also failed to reflect the drop in accuracy caused by mismatched telephone and studio recordings (McDougall et al., 2015). The research supports neither treating a confident voice identification as reliable on that basis alone nor treating confidence as meaningless; it supports recording confidence at the first identification and weighing it with the listening conditions.
Voice Parades: Procedures and Recent Research
Because unfamiliar-voice identification is fragile, procedure matters. Forensic phoneticians warned in the 1990s that voice lineups were becoming more common, were often run by people without the relevant expertise, and lacked published criteria (Hollien et al., 1995). Many safeguards developed for visual lineups translate to voices: an administrator who is unaware which voice is the suspect’s; an instruction that the voice may or may not be present; samples of similar length, recording quality, and speaking style; fillers similar to the suspect in accent, age, and general vocal qualities; neutral speech content rather than the words spoken during the crime; and a record of confidence at the time. In England and Wales, Home Office guidance from 2003 calls for nine samples of about one minute each, a design adapted from face-identification practice rather than tested on voices (Pautz et al., 2023). Having a witness listen to a single suspect speak raises the same concerns as a visual show-up.
Smith et al. (2020) found no accuracy difference between 15- and 30-second samples, and that a sequential procedure, deciding after each voice, was more accurate than a serial one, hearing all nine voices before deciding. Smith et al. (2023) then varied the pre-parade instruction: a strong warning that the voice might be absent, stressing the risk of convicting an innocent person, made listeners more cautious, reducing identifications whether or not the target was present; this time serial and sequential procedures did not reliably differ. Adjusting instructions thus offers a simple way to reduce false identifications, at some cost in correct ones. Pautz et al. (2023) found that 15-second samples performed about as well as 60-second samples and, given the low target-absent accuracy, recommended keeping nine-voice parades as added protection for a possibly innocent suspect. These studies used online listeners and brief recordings, but they show that system variables affect voice identification as they affect visual identification.
Memory for What Was Said
Earwitnesses are also asked what was said. Memory for conversation keeps the gist far better than the exact wording, which fades quickly and tends to be replaced by the listener’s interpretation. That interpretation depends on attention, context, and cues such as facial expression that a listener who could not see the speaker may have missed, and an impression that two people “were arguing” may outlast the words that produced it (Davis & Friedman, 2007). A witness may be accurate about the substance of a threat and mistaken about its exact words or about which speaker said what. These issues are treated in the Memory for Oral Communications paper.
Lay Listeners and Forensic Voice Comparison
In forensic voice comparison, a specialist compares a recording of an unknown speaker with one of a known speaker using auditory, acoustic-phonetic, or automatic methods. The older aural-spectrographic or “voiceprint” method produced divided court decisions and was criticized for lack of validation. Researchers have argued for methods that report a likelihood ratio (how probable the acoustic evidence is if one speaker produced both recordings, compared with different speakers) and that are tested under conditions reflecting the case (Morrison, 2009); an international group has published consensus recommendations for validating and reporting such systems (Morrison et al., 2021).
In a study using recordings that reflected an actual case, all 169 lay listeners performed worse than an automatic forensic-voice-comparison system at judging whether pairs of recordings came from the same speaker, and listeners familiar with the speaker’s language and accent did best (Basu et al., 2022). The study involved unfamiliar speakers and was conducted by researchers who develop such systems; it does not show that a witness who knows a speaker well is less accurate than a machine, but it suggests that a juror’s own comparison of a recording with a defendant’s voice should not be assumed accurate.
Limits of the Evidence
Earwitness research is far smaller than eyewitness research, and many studies are small. Most use brief recorded voices heard in calm conditions, student or online listeners, short delays, and no consequences; real earwitnesses may have been frightened, heard the voice amid noise, or known the speaker in ways no experiment reproduces. Familiar-voice studies are few, and results on serial versus sequential presentation and sample duration differ across studies. Group findings describe how often listeners err under defined conditions; they cannot establish whether a particular witness correctly identified a particular voice.
Questions to Ask About This Research
When Smith et al. (2020) is cited for 39% correct identifications and 6% correct rejections, counsel should ask who was studied and how: listeners heard a stranger’s recorded voice once and chose from nine recordings, conditions that may or may not resemble the case. When familiarity is relied on, how much was there, and might the speaker have whispered or spoken differently? Yarmey et al. (2001) found accuracy depended on both. When McDougall et al. (2015) is cited for poor telephone performance, did the case involve the same mismatch between hearing and testing? When Basu et al. (2022) is cited, was the system used in this case validated under matching conditions (Morrison et al., 2021)? And when Olsson et al. (1998) is offered to discount an earwitness’s certainty, counsel should remember that overconfidence across a group of listeners cannot establish that this listener’s identification is mistaken. These questions are explained in “Reading Research Claims: A Guide for Attorneys” on this site.
Earwitness Memory and the Law
Federal and Military Standards
Federal Rule of Evidence 901(b)(5) permits a witness to identify a voice heard at any time, firsthand or through transmission or recording, in circumstances linking it to the alleged speaker; the advisory committee note says familiarity may be acquired after the speaking at issue. A witness need only be minimally familiar with the voice, and weight is then for the jury (United States v. Ortiz, 2015); the rule addresses authentication, not reliability. Suggestive identification procedures are reviewed under Neil v. Biggers (1972) and Manson v. Brathwaite (1977), limited by Perry v. New Hampshire (2012) to suggestion arranged by law enforcement. A suspect may be required to speak for identification; neither speaking at a lineup (United States v. Wade, 1967) nor giving a voice exemplar (United States v. Dionisio, 1973) is testimonial. Expert testimony on earwitness memory or voice comparison is governed by Rule 702, as amended December 1, 2023, Daubert v. Merrell Dow Pharmaceuticals, Inc. (1993), and Kumho Tire Co. v. Carmichael (1999), which extended the gatekeeping obligation to all expert testimony, including technical or other specialized knowledge. In United States v. Angleton (2003), a district court granted the government’s motion to exclude a defense expert’s aural-spectrographic voice identification testimony after reviewing the science and protocols and finding it unreliable under Rule 702. In a separate opinion in the same case, the court admitted expert testimony on unintuitive earwitness factors (sample length, discussions among earwitnesses, and a single recording versus a voice lineup) and excluded testimony on intuitive ones.
In courts-martial, Military Rule of Evidence 901(b)(5) is worded the same as the federal rule, and the Court of Appeals for the Armed Forces has applied a flexible approach to authenticating recorded calls identified by the complainant and others familiar with the accused’s voice (United States v. Blanchard, 1998). Military Rule of Evidence 321 governs identifications of the accused and treats an identification process as unlawful if it is so suggestive as to create a substantial likelihood of misidentification. Its text neither mentions voices nor limits itself to visual identification, and no decision of the Court of Appeals for the Armed Forces applying it to a voice identification was located. Voice samples are not protected by the privilege against self-incrimination (United States v. Akgun, 1987). Expert testimony is assessed under Military Rule of Evidence 702 and United States v. Houser (1993); through Military Rule of Evidence 1102, the 2023 federal amendment became part of the military rule in 2025 (Joint Service Committee on Military Justice, 2025), and practitioners should confirm current text.
Arizona
Arizona Rule of Evidence 901(b)(5) is worded the same as the federal rule, and Arizona courts review challenged pretrial voice identifications under the framework used for visual ones. Under State v. Dessureault (1969), the State must prove by clear and convincing evidence that a challenged pretrial identification was not unduly suggestive or, if it was, that an in-court identification has an independent, reliable basis. In State v. Flynn (1973), a victim identified the defendant by sight and, less certainly, by voice at a lineup, and the court held that the State had shown at the Dessureault hearing that the lineup was not impermissibly suggestive. In State v. Goudeau (2016), a victim who had not seen her assailant listened to recordings of the suspect and five officers reading the same sentences; the suspect’s recording had an echo the others lacked, but the court found the procedure not impermissibly suggestive, noting the detective’s warning that recordings might vary and the victim’s half-hour exposure to the voice. The Arizona identification instruction (RAJI Standard Criminal 45) lists the witness’s “opportunity to view” among its factors and ends with any other factor affecting reliability, which leaves room to address listening conditions when the identification is by voice. Arizona Rule of Evidence 702 adopted the Daubert standard effective January 1, 2012, and was amended effective January 1, 2024, to track the 2023 federal amendment. State v. Chapple (1983), Arizona’s leading case on eyewitness expert testimony, held exclusion an abuse of discretion on its facts; admission otherwise remains within the trial court’s discretion (State v. Nordstrom, 2001).
Special Issues in Legal Practice
The principles described above apply with equal force to every party in a legal proceeding. The science operates the same way regardless of who is asking the questions, and the research summarized here does not by itself favor either side. What differs between the parties is not the science but the vantage point — the particular places where these issues tend to surface, and the particular errors each party is best positioned to guard against. The two summaries below identify those recurring issues. They are offered for orientation, not as litigation strategy; in every case the controlling question is what the evidence in that case will, and will not, support, and the expert’s role is to inform the trier of fact rather than to advocate for either side.
Special Issues for Prosecutors
The research supports several propositions prosecutors can legitimately present. Familiar voices are often recognized well, and a witness who knew the speaker and heard an extended, undisguised exposure is in a much stronger position than a stranger who heard a few words. Accuracy can be substantial under favorable conditions, and a thin verbal description does not show that the witness cannot recognize the voice, because voices are hard to describe. Prosecutors are well positioned to protect an earwitness identification by ensuring a properly constructed voice lineup, with the safeguards described above, rather than a single-voice confrontation, and by recording the witness’s confidence at the first identification.
The research also counsels caution. Unfamiliar-voice identifications are generally less accurate than face identifications, target-absent performance in several recent online studies was at or near chance, and confidence has been an inconsistent guide for voices across studies, so testimony that relies on a witness’s certainty alone goes beyond what the research supports. Whispering, tone changes, accent differences, telephone transmission, and delay all reduce accuracy, and an identification made after the witness learned who the suspect was carries the risks the Lindbergh case illustrates. One study of 169 lay listeners found they performed worse than an automatic system on unfamiliar speakers, and jurors’ own comparisons have not been shown to be accurate; expert voice comparison must meet Rule 702.
Special Issues for Defense
The research supports several propositions defense counsel can legitimately present. Identification of a voice heard once and briefly is error-prone, and the conditions that reduce accuracy, including short exposure, delay, disguise, emotional tone, other-accent or other-language speech, and mismatched recording quality, are well documented. Single-voice confrontations, a suspect recording that stands out, use of the words spoken during the crime, and the absence of a warning that the voice may not be present are departures from recommended practice that warrant scrutiny. Expert testimony on these principles, with evidence of the actual listening and testing conditions, is a legitimate way to help the factfinder weigh an earwitness identification.
The research also counsels caution. Voices can be recognized, and familiar voices often are; the research does not show that voice identification is inherently unreliable or that a particular identification is wrong. An expert may describe the research but may not tell the jury whether the witness is correct. Some procedural findings, such as the advantage of sequential presentation, have not replicated consistently and should be stated with that limit. Forensic voice comparison that favors the defense is subject to the same validation demands as when it favors the prosecution.
This paper is for educational purposes only and is not legal advice. The author is a psychologist, not an attorney; summaries of law are general and must be verified against current authority or with a licensed attorney.
References
Basu, N., Bali, A. S., Weber, P., Rosas-Aguilar, C., Edmond, G., Martire, K. A., & Morrison, G. S. (2022). Speaker identification in courtroom contexts – Part I: Individual listeners compared to forensic voice comparison based on automatic-speaker-recognition technology. Forensic Science International, 341, Article 111499. https://doi.org/10.1016/j.forsciint.2022.111499
Clifford, B. R. (1980). Voice identification by human listeners: On earwitness reliability. Law and Human Behavior, 4(4), 373–394. https://doi.org/10.1007/BF01040628
Davis, D., & Friedman, R. D. (2007). Memory for conversation: The orphan child of witness memory researchers. In M. P. Toglia, J. D. Read, D. F. Ross, & R. C. L. Lindsay (Eds.), The handbook of eyewitness psychology: Vol. 1. Memory for events (pp. 3–52). Lawrence Erlbaum Associates.
Goggin, J. P., Thompson, C. P., Strube, G., & Simental, L. R. (1991). The role of language familiarity in voice identification. Memory & Cognition, 19(5), 448–458. https://doi.org/10.3758/BF03199567
Hollien, H., Huntley, R., Künzel, H., & Hollien, P. A. (1995). Criteria for earwitness lineups. Forensic Linguistics, 2(2), 143–153. https://doi.org/10.1558/ijsll.v2i2.143
Joint Service Committee on Military Justice. (2025). Amendments to the Military Rules of Evidence since 1984 (G. E. Maggs, Ed.; 2025 ed.). https://jsc.defense.gov/Portals/99/MRE%20Amendments%202025.pdf
Lavan, N., Burton, A. M., Scott, S. K., & McGettigan, C. (2019). Flexible voices: Identity perception from variable vocal signals. Psychonomic Bulletin & Review, 26(1), 90–102. https://doi.org/10.3758/s13423-018-1497-7
McDougall, K., Nolan, F., & Hudson, T. (2015). Telephone transmission and earwitnesses: Performance on voice parades controlled for voice similarity. Phonetica, 72(4), 257–272. https://doi.org/10.1159/000439385
McGehee, F. (1937). The reliability of the identification of the human voice. The Journal of General Psychology, 17(2), 249–271. https://doi.org/10.1080/00221309.1937.9917999
Morrison, G. S. (2009). Forensic voice comparison and the paradigm shift. Science & Justice, 49(4), 298–308. https://doi.org/10.1016/j.scijus.2009.09.002
Morrison, G. S., Enzinger, E., Hughes, V., Jessen, M., Meuwly, D., Neumann, C., Planting, S., Thompson, W. C., van der Vloed, D., Ypma, R. J. F., Zhang, C., Anonymous, A., & Anonymous, B. (2021). Consensus on validation of forensic voice comparison. Science & Justice, 61(3), 299–309. https://doi.org/10.1016/j.scijus.2021.02.002
Olsson, N., Juslin, P., & Winman, A. (1998). Realism of confidence in earwitness versus eyewitness identification. Journal of Experimental Psychology: Applied, 4(2), 101–118. https://doi.org/10.1037/1076-898X.4.2.101
Orchard, T. L., & Yarmey, A. D. (1995). The effects of whispers, voice-sample duration, and voice distinctiveness on criminal speaker identification. Applied Cognitive Psychology, 9(3), 249–260. https://doi.org/10.1002/acp.2350090306
Pautz, N., McDougall, K., Mueller-Johnson, K., Nolan, F., Paver, A., & Smith, H. M. J. (2023). Identifying unfamiliar voices: Examining the system variables of sample duration and parade size. Quarterly Journal of Experimental Psychology, 76(12), 2804–2822. https://doi.org/10.1177/17470218231155738
Philippon, A. C., Cherryman, J., Bull, R., & Vrij, A. (2007). Earwitness identification performance: The effect of language, target, deliberate strategies and indirect measures. Applied Cognitive Psychology, 21(4), 539–550. https://doi.org/10.1002/acp.1296
Saslove, H., & Yarmey, A. D. (1980). Long-term auditory memory: Speaker identification. Journal of Applied Psychology, 65(1), 111–116. https://doi.org/10.1037/0021-9010.65.1.111
Smith, H. M. J., Bird, K., Roeser, J., Robson, J., Braber, N., Wright, D., & Stacey, P. C. (2020). Voice parade procedures: Optimising witness performance. Memory, 28(1), 2–17. https://doi.org/10.1080/09658211.2019.1673427
Smith, H. M. J., Roeser, J., Pautz, N., Davis, J. P., Robson, J., Wright, D., Braber, N., & Stacey, P. C. (2023). Evaluating earwitness identification procedures: Adapting pre-parade instructions and parade procedure. Memory, 31(1), 147–161. https://doi.org/10.1080/09658211.2022.2129065
Stevenage, S. V. (2018). Drawing a distinction between familiar and unfamiliar voice processing: A review of neuropsychological, clinical and empirical findings. Neuropsychologia, 116(Pt B), 162–178. https://doi.org/10.1016/j.neuropsychologia.2017.07.005
Stevenage, S. V., Clarke, G., & McNeill, A. (2012). The “other-accent” effect in voice recognition. Journal of Cognitive Psychology, 24(6), 647–653. https://doi.org/10.1080/20445911.2012.675321
Stevenage, S. V., Howland, A., & Tippelt, A. (2011). Interference in eyewitness and earwitness recognition. Applied Cognitive Psychology, 25(1), 112–118. https://doi.org/10.1002/acp.1649
Yarmey, A. D. (1995). Earwitness speaker identification. Psychology, Public Policy, and Law, 1(4), 792–816. https://doi.org/10.1037/1076-8971.1.4.792
Yarmey, A. D., Yarmey, A. L., Yarmey, M. J., & Parliament, L. (2001). Commonsense beliefs and the identification of familiar voices. Applied Cognitive Psychology, 15(3), 283–299. https://doi.org/10.1002/acp.702
Legal Authorities
Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993).
Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999).
Manson v. Brathwaite, 432 U.S. 98 (1977).
Neil v. Biggers, 409 U.S. 188 (1972).
Perry v. New Hampshire, 565 U.S. 228 (2012).
State v. Chapple, 135 Ariz. 281, 660 P.2d 1208 (1983).
State v. Dessureault, 104 Ariz. 380, 453 P.2d 951 (1969).
State v. Flynn, 109 Ariz. 545, 514 P.2d 466 (1973).
State v. Goudeau, 239 Ariz. 421, 372 P.3d 945 (2016).
State v. Nordstrom, 200 Ariz. 229, 25 P.3d 717 (2001).
United States v. Akgun, 24 M.J. 434 (C.M.A. 1987).
United States v. Angleton, 269 F. Supp. 2d 868 (S.D. Tex. 2003) (earwitness-memory expert testimony), and 269 F. Supp. 2d 892 (S.D. Tex. 2003) (aural spectrography).
United States v. Blanchard, 48 M.J. 306 (C.A.A.F. 1998).
United States v. Dionisio, 410 U.S. 1 (1973).
United States v. Houser, 36 M.J. 392 (C.M.A. 1993).
United States v. Ortiz, 776 F.3d 1042 (9th Cir. 2015).
United States v. Wade, 388 U.S. 218 (1967).
Fed. R. Evid. 702.
Fed. R. Evid. 901(b)(5).
Ariz. R. Evid. 702.
Ariz. R. Evid. 901(b)(5).
Mil. R. Evid. 321.
Mil. R. Evid. 702.
Mil. R. Evid. 901(b)(5).
Mil. R. Evid. 1102.
Revised Arizona Jury Instructions (Criminal), Standard Criminal 45 (Identification).
Related topic papers: Memory for Oral Communications · Eyewitness Memory · Lineup Construction · Show-Ups · Confidence and Accuracy · The Cross-Race Effect
Related topics from Arizona Forensic Services: Courts-Martial