OPINION OF THE COURT
Bracken, P. J.The People assert that the defendant, John Scott, was one of two masked men who robbed a grocery in Queens on March 21, 1996. Yet of the several persons who were present in the grocery at the time of the robbery, the People produced none who, at trial, identified the defendant as one of the robbers. There is no proof that the defendant ever admitted to having participated in the robbery, and there is no physical evidence linking him to that crime. The only evidence linking the defendant to the crime consists in the proof of a single extrajudicial identification made by one witness based not on the defendant’s appearance, but based instead on the defendant having spoken a few words in a voice which the identifying witness thought matched that of the taller of the two robbers. In our opinion, the convictions are against the weight of the evidence and, therefore, in the exercise of our factual review power, we reverse the judgment of conviction and dismiss the indictment.
The chief witness for the People was William Vizcarrondo, who worked in the grocery where the robbery occurred. As alluded to above, and as will be elaborated upon below, Vizcarrondo’s trial testimony concerning his extrajudicial voice-based identification of the defendant constitutes the cornerstone of the People’s case. Yet Vizcarrondo was not produced at the pretrial Wade hearing which was conducted in this case in order to permit the People to meet their burden of proving that the extrajudicial voice-based identification made by their principal witness was untainted by any suggestiveness.
In order to meet their burden at the Wade hearing, the People called one witness, Detective Gary Russo, who had been assigned to investigate the robbery. Detective Russo testified that, several days after the robbery, Vizcarrondo was asked to view a lineup which included the defendant. Vizcarrondo initially viewed the six participants in the lineup while they were arrayed in a seated position. He then asked Detective Russo to have them stand, and to have each of the participants *100utter the phrase “[R]un the shine,” a phrase which the taller of the two robbers had used during the robbery. It was only after the participants had each uttered these words that Vizcarrondo identified the defendant.
Following the Wade hearing, the Supreme Court denied that branch of the defendant’s motion which was to suppress the use of any evidence relating to Vizcarrondo’s extrajudicial identification. The Supreme Court stated, in its decision and order, “[t]here was nothing elicited in the testimony [at the Wade hearing] to indicate that the voice identification was suggestive in any way.”
At trial, Vizcarrondo testified that at approximately 8:00 p.m. on March 21, 1996, two men, both wearing black jeans, black leather jackets, black ski masks, and knit hats, and both brandishing guns, burst into his father’s grocery and announced a robbery.
Vizcarrondo described the taller of the two men, that is, the person who the People now theorize was the defendant, as being approximately six feet and one inch tall, and weighing approximately 170 pounds. The taller man had a voice which Vizcarrondo described as deep, and which he also described as being a little different than any he had ever heard. Vizcarrondo was also able to observe the color of the portion of the taller man’s skin which was exposed by a small opening in the mask in the area of his eyes. Based on this, he testified that the taller of the two robbers was a light-skinned black man.
During the robbery, the taller man asked Vizcarrondo to produce the.cash in the cash register, and to “run the shine.” Vizcarrondo took this as being a request to turn over his jewelry. Vizcarrondo surrendered the cash, as well as his chain and ring. The taller robber told him that he could keep his wedding band. Before leaving the grocery with his accomplice, the taller robber asked about the score of a basketball game, and asked for cigarettes and a match. The entire incident lasted approximately four minutes.
At trial, Vizcarrondo testified that he could not identify the defendant as the taller of the two robbers. He stated that “too much time [had] passed.” He also conceded that such an identification was impossible because the taller of the two robbers had been masked. He testified only that, at the lineup conducted several days after the robbery, he pointed to the person who had the same voice as the taller of the two robbers.
Questioned about the circumstances surrounding his prior, extrajudicial identification of the defendant, Vizcarrondo testi*101fied that the defendant was one of two men whose skin tone was noticeably lighter than that of the other participants in the lineup. He also testified that he did not select any of the lineup participants until after each of the participants in the lineup had repeated the phrase “[g]et on the floor, and run the shine.” In describing the similarity between the voice of the defendant as it sounded at the time of the lineup and the voice of the taller robber, Vizcarrondo testified that both were “deep” voices.
At trial, the People produced, as witnesses, other individuals who were present in the grocery during the course of the robbery. None of these witnesses identified the defendant. Their testimony tends to confirm the People’s theory that one of the two robbers had a voice which was “rough,” “deep,” “cool,” “commanding,” or “hard.”
At trial, the police also produced Detective Russo, who, as mentioned above, had conducted the lineup which included the defendant. The prosecutor elicited from this witness the fact that the lineup was conducted after he had spoken to an individual named Eric Mcllwain, who was not called as a witness.
On appeal, the defendant contends, inter alia, that the Supreme Court erred in failing to suppress Vizcarrondo’s testimony with regard to the pretrial voice identification of the defendant, the only competent evidence of guilt in this case. In this regard, the defendant asserts that the People failed to meet their burden of proving that the voice-based lineup identification was not unduly suggestive (see, People v McRae, 195 AD2d 180; People v Shepard, 162 AD2d 226; People v Wong, 133 AD2d 184). Next, he argues that the jury verdict is against the weight of the evidence. We agree with the defendant that the jury verdict is against the weight of the evidence. Therefore, the defendant’s argument relating to the admissibility of the pretrial voice identification, and the defendant’s remaining arguments as well, are academic and need not be addressed.
“It is a fundamental rule that due process of law protects an accused from conviction ‘except upon proof beyond a reasonable doubt of every fact necessary to constitute the crime with which he is charged’ ” (People v Egan, 72 AD2d 239, 242, quoting Matter of Winship, 397 US 358, 364). Referring to cases in which victims of crimes committed by strangers make later visual identifications in court, the Supreme Court stated, in United States v Wade (388 US 218, 228-229), that “[t]he identification of strangers is proverbially untrustworthy * * * ‘[t]he influence of improper suggestion upon identifying wit-< *102nesses probably accounts for more miscarriages of justice than any other single factor — perhaps it is responsible for more such errors than all other factors combined.’ ” The defendant contends, on appeal, that an “ear-witness” is, if anything, less reliable than an eyewitness, and that the evidence produced in this case did not establish his guilt beyond a reasonable doubt. Without delving into the relative reliability of identifications based on voice, as opposed to identifications based on other physical features (but see, People v Collins, 60 NY2d 214, 218 [voice identifications possibly less reliable]), we find that the People failed to produce enough evidence to establish the defendant’s identity as one of the perpetrators of the robbery in this instance.
The chief witness for the People in this case made no visual identification of the defendant at trial, and conceded his inability to do so. This witness also made no in-court identification of the defendant based on his presumed ability to recall the supposedly distinctive voice of the tall robber, because the defendant was not compelled to speak at trial, and no recording of his voice was offered into evidence. Such voice exemplars are admissible (see, People v Berg, 239 AD2d 97; People v Diaz, 196 AD2d 790; People v Forant, 142 AD2d 891). Had such an exemplar been admitted into evidence, the jury would at least have had both the opportunity to hear for itself the supposedly distinctive character of the defendant’s voice and, consequently, the opportunity to assess the reliability of the claim that the complaining witness was able to distinguish that voice from all others.
As the result of the absence of any such voice exemplar as part of the record on appeal, this Court is now in the same position as that in which the jurors found themselves at the time of trial. While we do have testimony of several witnesses who characterized the tall robber’s voice as “rough,” “deep,” “hard,” etc., we note that these witnesses (unlike Vizcarrondo) were apparently never offered the opportunity to compare this supposedly distinctive voice with the voice of the defendant. We thus have nothing more than the opinion offered by a single witness, Vizcarrondo, that this voice and the voice of the defendant were one and the same.
The People argue, in their brief, that Vizcarrondo’s lineup identification was based on his observation of the physical attributes displayed by the defendant at the time of the lineup, as well as upon his voice. The People argue accordingly that the reliability of this lineup identification is bolstered by *103certain proof which tended to show that the robbery occurred under bright lights, which gave Vizcarrondo an “ample opportunity” to “observe the defendant’s complexion and height.” However, the weight of the evidence supports the conclusion that Vizcarrondo would not have been able to make an identification based on factors other than the supposed similarity of the defendant’s voice and that of the taller perpetrator. This is reflected in the following testimony, elicited from Vizcarrondo on cross-examination:
“Q In any event, you testified — you said that you recognized the person that robbed you in part by his skin color and height, correct sir?
“A Correct.
“Q I want you to look around this room and look over there, look very carefully. Do you recognize this person as the person who robbed you?
“A I recognize him from the lineup, but I don’t recognize the person that robbed me because I couldn’t see because of the mask.
“Q Right. But you said you recognized him by his skin color and height. And you’re saying you recognized him at the lineup. But looking at him here — take a good look at him, sir. Right now, July 14, ‘97, you can’t say it’s the guy that robbed you, right?
“A I can’t say that.
“Q You’re looking at his skin color, right?
“A Yes.
“Q You’re looking at the greater half of his body, right? Can’t recognize him, can you?
“A No.
“Q And connect him with this robbery?
“A No.
“Q So doesn’t it come down to this, that you’re looking at his skin color, you’re looking at him right here, and you’re saying it’s not enough. And the only way you can connect him is with the lineup when you heard someone speak, correct?
“A Correct.
“Q Thank you.”
The defendant is correct, then, in arguing that the judgment of conviction rests entirely on Vizcarrondo’s voice-based identification of the defendant from a police-arranged lineup. Even assuming, without deciding, that the testimony relating *104to this out-of-court procedure was properly admitted into evidence, the fact remains that the jury had no opportunity to hear the defendant’s voice, or to assess its allegedly distinctive quality. Furthermore, the jury had no opportunity to hear the voices of the five other members of the lineup from which the defendant was selected. Had the People proved that their principal witness selected the defendant from a lineup consisting of other deep-voiced men, then the weight to be afforded to any evidence relating to such identification would certainly be stronger than if it were to appear that the defendant was instead chosen from a lineup in which every other participant had a voice noticeably higher than his own. As it stands, the record is silent as to the reliability of the voice-identification, and, aside from whether the silence of the record in this respect would warrant suppression of any evidence relating to the lineup identification as a matter of due process (see, United States v Wade, supra; People v McRae, supra), we find that, in the circumstances of this case, the People’s failure, at the Wade hearing, to demonstrate more clearly the reliability of the identification procedure employed by the police resulted in a distinct failure of proof at the time of the later trial.
In sum, the verdict is against the weight of the evidence, and the judgment must therefore be reversed, the indictment is dismissed, and the matter is remitted to the Supreme Court, Queens County, for the purpose of entering an order in its discretion pursuant to CPL 160.50.
Altman, J. P., Goldstein and McGinity, JJ., concur.
Ordered that the judgment is reversed, on the facts, the indictment is dismissed, and the matter is remitted to the Supreme Court, Queens County, for the purpose of entering an order in its discretion pursuant to CPL 160.50.