Robert William Schultz, Jr., was convicted by a jury of attempting to extort funds from the Zapp National Bank of St. Cloud, Minnesota, in violation of 18 U.S.C. § 1951(a) (1976). The District Court ^[1] sentenced him to five years’ imprisonment. On appeal Schultz contends that (1) the government impermissibly cross-examined and commented on his failure to produce a supporting witness, (2) the government impermissibly commented on his post-arrest silence, and (3) a voice-exemplar “lineup” was so suggestive as to give rise to a very substantial likelihood of irreparable misidentification. We reject Schultz’s contentions and affirm.

I.

On April 14, 1982, Tracy Morse, a receptionist at the Zapp National Bank, received a telephone call from a man who did not identify himself. She connected the caller with Roger Poganski, an assistant vice-president of the bank. The caller told Mr. Poganski to place $20,000 at the base of a Conoco sign on the other side of town within thirty minutes or hand grenades would be thrown at the bank. Mr. Poganski put together a “dummy package,” drove to the drop site, and placed the package under the Conoco sign. Law enforcement officers, who had the area under surveillance, observed Schultz stop his car beside the Conoco sign and then drive away. Schultz was arrested shortly thereafter, and the dummy package was found in his car. After Schultz was informed of his rights, he made a statement to the police. He admitted picking up the package but denied making any telephone calls to the bank.

II.

At trial, Schultz testified that he was in the area of the Conoco station to meet a friend, whom he identified for the first time as Bill Anderson, so that they[*367] could shop for machinery. The government cross-examined Schultz about why he had not called Anderson as a witness and emphasized the point in its closing argument. Schultz contends that this shifted the burden to him to convince the jury that he was innocent. We disagree. The prosecutor is free to comment on the failure of the defendant to call an available alibi witness. E.g., Yancey v. Housewright, 664 F.2d 187, 190 (8th Cir.1981) (dictum). Although Schultz claimed during cross-examination that he “wouldn’t have the vaguest idea where to look for [Anderson],” T. 211, ^[2] the prosecutor cast doubt on the truth of this assertion by continued cross-examination, and the jury was free to disbelieve the defendant. Moreover, the prosecutor told the jury during his closing argument that Schultz had no duty to produce evidence, T. 278, 296, and we presume that the court similarly instructed the jury. In these circumstances, we find no error.

III.

At trial, the government cross-examined Schultz about his failure to tell law-enforcement officers after his arrest about his plan to meet Bill Anderson. Schultz argues that the government is prohibited from impeaching a defendant with his post-arrest silence. This is not a case, however, in which the defendant was completely silent after his arrest. After being given satisfactory Miranda warnings, Schultz chose to make a statement about why he was in the area of the drop site: he said he was shopping for machinery. Therefore, when he chose to testify at trial, he was subject to cross-examination with respect to his prior, arguably inconsistent statement concerning the same subject matter. See Anderson v. Charles, 447 U.S. 404, 100 S.Ct. 2180, 65 L.Ed.2d 222 (1980).

IV.

On July 12, 1982, the government made a telephone recording of Schultz’s voice while Schultz read from a prepared script. On July 14 the government recorded the voices of five retired FBI agents while they read from the same script. All the voices were recorded on the same recorder and on the same telephone. During the “lineup,” which was held on July 16, Ms. Morse and Mr. Poganski were separated and were required to listen to the voices individually. Neither Ms. Morse nor Mr. Poganski was told that the defendant’s voice was among those played. The defendant’s counsel was present during the lineup, and the order in which the tapes were played was changed at her request. T. 110. Both Ms. Morse and Mr. Poganski identified Schultz’s voice as that of the extortionist.

Schultz contends that this voice-exemplar lineup was “so impermissibly suggestive as to give rise to a very substantial likelihood of irreparable misidentification,” Simmons v. United States, 390 U.S. 377, 384, 88 S.Ct. 967, 971, 19 L.Ed.2d 1247 (1968), that Ms. Morse’s and Mr. Poganski’s testimony about the lineup and identification of Schultz as the caller should have been suppressed. He argues that his voice was halting and uneducated with a peculiar accent, while the five FBI agents’ voices were educated and highly trained. We have carefully reviewed the recordings, and we do not believe that the difference in accent and vocal style between Schultz’s voice and the other five voices was so pronounced as to amount to impermissible suggestiveness. As we observed in United States v. Lewis, 547 F.2d 1030 (8th Cir.1976), cert. denied, 429 U.S. 1111, 97 S.Ct. 1149, 51 L.Ed.2d 566 (1977), “Police stations are not theatrical casting offices; a reasonable effort to harmonize the lineup is normally all that is required.” Id. at 1035.

However, in reviewing the tapes, we found a marked difference in volume; the recordings of the five FBI agents’ voices were loud and had good sound resolution, while the recording of the defendant’s voice was soft, and there was a good deal of static or “tape hiss.” We are not sure how to explain this discrepancy. At trial de[*368] fense counsel asked the agent who set up the taping how he explained the difference in quality of the tapes, and he testified he had not noticed any difference in quality. T. 114. Mr. Poganski stated, “He [the caller] spoke very softly on the phone. When we originally listened to the tapes they had it up. I had them turn it down to about the same level that it was on the phone.” T. 29. Perhaps the difference in sound level and tape quality is completely innocent. We do know that the difference is pronounced, at least in the tapes’ present condition, and if the difference was so great at the time of the lineup, there was certainly an impermissible level of suggestiveness. We therefore, in the alternative, proceed to the second criterion of the test for admissibility. If the lineup was impermissibly suggestive, are the identifications nevertheless admissible because there was no very substantial likelihood of irreparable misidentification?

We conclude that there was no very substantial likelihood of irreparable misidentification, and thus the conviction must be affirmed. Ms. Morse, although she spoke with the extortionist for less than a minute, testified that she took a particular interest in voices of persons who telephoned the bank, because she liked to get to know the bank’s customers’ voices. T. 5. She spoke with the caller longer than she would with a normal customer. T. 8. Mr. Poganski spoke with the extortionist for three or four minutes, and he testified that he listened very closely to the voice so that he could identify it. T. 13. He made the caller repeat his instructions several times. Although there is no evidence that either witness described the extortionist’s voice prior to trial, both witnesses identified Schultz’s voice as that of the extortionist, and Mr. Poganski’s identification was definite. T. 25. Finally, we cannot say that the three-month interval between the incident and the lineup in and of itself rendered the identifications unreliable. Under these circumstances, the reliability of the witnesses’ identification of Schultz as the extortionist was a question for the jury to determine.

Accordingly, the judgment of conviction is affirmed.