United States v. Drones, 218 F.3d 496

218 F.3d 496

5th Cir.

2000

EMILIO M. GARZA, Circuit Judge:

Appellee Janadriek Kemont Drones was convicted of (1) conspiracy to possess crack cocaine with intent to distribute and (2) aiding and abetting possession of crack cocaine. Drones petitioned for federal ha-beas corpus relief claiming, inter alia, ineffective assistance of counsel. After conducting an evidentiary hearing, the district court granted habeas relief, holding that Drones’s trial counsel was ineffective in failing to investigate and present voice identification evidence. The district court then vacated Drones’s convictions and ordered a new trial. The United States of America (the “government”) now appeals the district court’s ruling, arguing that Drones received effective assistance of counsel. For the reason set forth below, we reverse the judgment of the district court.

On April 25, 1995, Sergeant Gregory Haire of the Texas Department of Public Safety (“DPS”) met with two confidential informants (“CIs”) about arranging a purchase of crack cocaine. Pursuant to Haire’s request, the CIs contacted two individuals whom they had identified as possible drug dealers. "When one of the individuals returned the call, DPS recorded the telephone conversation as well as a second call from the same individual made later the same day. During the course of these conversations, the CIs planned to meet the caller at a Burger King restaurant to complete the drug transaction.

Later that afternoon, Haire and the CIs went to the parking lot of the Burger King where they saw a parked Ford Mustang. "When the CIs approached the vehicle, the occupants of the car — Vernon and Arnold Freddie (the “Freddies”) and Drones — let them into the car and began to weigh the drugs. When Haire signaled for arrest, nearby officers surrounded the car and arrested the Freddies as well as Drones, who was by then standing outside of the car. Drones and the Freddies were charged with (1) conspiring to possess cocaine base with intent to distribute, (2) aiding and abetting in the possession of cocaine base with intent to distribute, and (3) aiding and abetting in the use and carrying of a firearm in relation to a drug trafficking offense.

Drones was tried with Vernon Freddie. 1 Before the case was submitted to the jury, the trial court granted Drones’s motion for judgment of acquittal on the firearm count but denied Drones’s motion with respect to the other two counts. At trial, Haire testified that Drones was sitting in the driver’s seat during part of the drug transaction. At the time of the arrest, Haire stated, Drones was standing outside of the car and “made a motion as if to attempt to run” when approached by the police. When the police searched the vehicle, they found packages of cocaine in the front passenger compartment.

While Haire was on the witness stand, the government tapes were played for the jury, and Haire testified that he recognized Drones’s voice as the “unknown” speaker on the tapes. On cross-examination of Haire, Drones’s court-appointed counsel, Christopher Goldsmith, emphasized the sloppy nature of the police investigation and the fact that, despite the high reliability of voice identification evidence, the government had put forth no expert testimony identifying Drones’s voice as the unknown voice on the tapes. Goldsmith presented no expert voice identification evidence of his own. 2

Goldsmith called several witnesses to testify in Drones’s defense. First, the Freddies both testified that Drones did not participate in the drug transaction. Arnold Freddie further testified that he was involved in the taped phone conversations, and that he did not recognize Drones’s voice on the government tapes. He stated that the first time he met Drones was in the Burger King parking lot.

Next, Krisna Brown-Drones (“Krisna”), Drones’s then-fiancee, testified that she was with Drones on the day of the arrest. More specifically, she testified that she and Drones were at her sister’s apartment, which did not have a phone, between noon and 4:30. At approximately 4:30, Krisna testified, Drones left the apartment to get something to eat.

The jury convicted Drones of both drug charges. Shortly thereafter, the trial court reconsidered and granted Drones’s motion for judgment of acquittal on the conspiracy and aiding and abetting claims. In reversing its prior ruling, the court found that (1) Drones’s voice was probably not on the tape, and (2) the government had presented insufficient evidence to convict Drones. The government appealed, and we reversed and remanded with instructions to the district court to reinstate the jury verdict. On remand, the district court reinstated the verdict and sentenced Drones to 210 months imprisonment. We affirmed the conviction and sentence.

Drones filed a timely motion to vacate, set aside, or correct his sentence pursuant to 28 U.S.C. § 2255 claiming, inter alia, that he was denied effective assistance of counsel. Specifically, Drones contended that Goldsmith had rendered ineffective assistance by failing to investigate and present expert and layperson voice identification testimony that his voice was not on the government tapes. The district court conducted an evidentiary hearing on the issue of “whether counsel’s failure to investigate exculpatory evidence regarding Drones’s identity rendered his performance constitutionally deficient.”

Drones, represented by newly appointed counsel, presented four witnesses at the evidentiary hearing. First, Steve Cain, a forensic scientist and voice identification expert, testified about the procedures used in voice analysis and his own analysis of the government tapes. 3 Cain testified that, after comparing the unknown voice on the government tapes with a voice exemplar obtained from Drones, he reached a finding of “probable elimination,” meaning that at least 80% of the comparable words in the samples were dissimilar aur ally and spectrographieally. 4

While Cain testified that there were published recommended procedures for conducting voice identification examinations, he also acknowledged several weaknesses in spectrographic analysis. Specifically, he testified that there was no set of objective criteria against which to check the accuracy of a particular expert’s analysis and that voice identification analysis was 'largely subjective in that the examiner ultimately decides whether two spectro-graphs match one another.

Krisna testified that she was familiar with Drones’s voice and therefore able to distinguish it from other voices. She further testified that Goldsmith never asked her to listen to the audiotapes prior to the trial. After listening to the government tapes at the evidentiary hearing, she testified that Drones’s voice was not on the government tapes.

Next, Lieutenant Colonel Dave Johnson, Drones’s grandfather, testified that he was also familiar with Drones’s voice, and that Goldsmith had never asked him to listen to the government tapes. After listening to the tapes at the hearing, he stated that he did not recognize Drones’s voice on the tapes. Johnson further testified that while he had suffered some hearing loss, that loss did not prevent him from recognizing Drones’s voice.

Goldsmith also testified at the evidentia-ry hearing. He stated that he considered the tape to be a “critical” piece of evidence at the trial because, without the identification of Drones’s voice, the government would have had a “much more difficult time” proving that Drones was part of a conspiracy. He also stated that “several week or months” before trial, Drones told him that it was not his voice on the tape. While Goldsmith listened to the tape several times, he was unable to determine whether or not Drones’s voice was on the tapes. Most importantly for the purposes of this appeal, Goldsmith testified that he made a “strategic decision” to base Drones’s defense upon sloppy police work, the alibi testimony of Brown, and the Freddies’ testimony, rather than upon expert voice identification testimony that Drones’s voice was not on the government tapes. Because of this decision, Goldsmith testified, he never sought out expert or layperson evaluations of the tapes.

Finally, the government presented the expert testimony of Bruce Koenig, a private consultant and former Federal Bureau of Investigation employee. 5 Koenig testified that very little research has been done in the area of “courtroom application of spectrographic voice identification,” largely because since the 1970’s, many researchers have felt that spectrographic comparison could not produce reliable results. He stated that “almost nobody” in the relevant scientific community uses spectrographic voice identification because there is no theoretical basis for the proposition that an individual’s voice is truly unique and identifiable. Nonetheless, Koenig, like Cain, found a “probable elimination” as to one of the government tapes. With respect to the other tape, Koenig used only “critical listening” and concluded that Drones’s voice was dissimilar to the voice on the government tapes.

After making extensive findings of fact and law, the district court held that Drones was denied effective assistance of counsel due to Goldsmith’s failure to investigate the identity of the voice on the government tapes, and it granted habeas relief. More specifically, the district court found that counsel did not make a “reasonable and informed choice to forego investigation of the tape after listening to the tape on his own” and that “by choosing not to investigate or present such evidence, Goldsmith denied the jury the opportunity to hear this potentially exculpatory evidence on [the issue of voice identification].” The government filed this timely appeal in which it contends that (1) defense counsel’s decision not to investigate the identity of the voice on the tape was a reasonable strategic legal decision, and (2) even if counsel’s performance was deficient, that deficiency did not prejudice Drones’s defense. We review a district court’s finding of fact on requests for habeas relief for clear error and its rulings on issues of law de novo. See Fairman v. Anderson, 188 F.3d 635, 640 (5th Cir.1999). Determinations regarding the effectiveness of counsel are mixed questions of law and fact which we review de novo. See Ransom v. Johnson, 126 F.3d 716, 721 (5th Cir.1997).

We review ineffective assistance claims under the familiar two-prong analysis set forth in Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). First, a petitioner must demonstrate that his attorney’s performance was deficient. See Strickland, 466 U.S. at 688, 104 S.Ct. 2052, 80 L.Ed.2d 674. In order to be deficient, counsel’s performance must be “outside the wide range of professionally competent assistance.” Id. at 690, 104 S.Ct. 2052, 80 L.Ed.2d 674. If he succeeds in satisfying the first hurdle, then a petitioner must also demonstrate that the deficient performance prejudiced the defense such that “there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Id. at 694, 104 S.Ct. 2052, 80 L.Ed.2d 674; see also Bryant v. Scott, 28 F.3d 1411 (5th Cir.1994).

Our review of counsel’s performance is “highly deferential,” and we “indulge a strong presumption that counsel’s performance falls within the wide range of reasonable professional assistance.” Strickland, 466 U.S. at 689, 104 S.Ct. 2052, 80 L.Ed.2d 674. Recognizing that in “failure to investigate” cases the temptation to rely on hindsight is strong, the Supreme Court stated that “strategic choices made after thorough investigation of law and facts relevant to plausible options are virtually unchallengeable.” Id. at 690, 104 S.Ct. 2052, 80 L.Ed.2d 674. However, “strategic choices made after less than complete investigation are reasonable precisely to the extent that reasonable professional judgments support the limitations on investigation. In other words, counsel has a duty to make reasonable investigations or to make a reasonable decision that makes particular investigations unnecessary.” Id. at 690-91, 104 S.Ct. 2052, 80 L.Ed.2d 674. Strickland does not require us to defer to decisions that are uninformed by an adequate investigation into the controlling facts and law. See Moore v. Johnson, 194 F.3d 586, 615 (5th Cir.1999).

There is little doubt that Goldsmith failed to make any, let alone a reasonable, investigation into the identity of the unknown voice on the government tapes. Goldsmith himself testified at the eviden-tiary hearing that he never even investigated the availability of expert voice identification analysis of the tapes. He further testified that he chose not to play the tapes for either Drones or members of his family. Both Krisna and Johnson testified that while Goldsmith conferred with them before trial, he never asked them to review the tapes. Accordingly, the question here is whether Goldsmith’s decision to limit his investigation so as to develop neither expert nor layperson identification testimony was reasonable under the circumstances. See Strickland, 466 U.S. at 691, 104 S.Ct. 2052, 80 L.Ed.2d 674; id. at 699, 104 S.Ct. 2052, 80 L.Ed.2d 674 (stating that when assessing counsel’s conduct, we must “reconstruct the circumstances of counsel’s challenged conduct and [evaluate] the conduct from counsel’s perspective at the time”).

As an initial matter, we note the significance of the government tapes to the prosecution’s case. Drones was charged with both aiding and abetting possession of crack cocaine and conspiracy to possess crack with an intent to distribute. Probative to both claims were the taped conversations in which the CIs planned to meet an unknown caller to complete a drug transaction. The government presented only one witness — Officer Haire — to testify that he recognized Drones’s voice on the government tapes. See Fed.R.Evid. 901(b)(5) (providing that “identification of a voice ... by opinion based upon hearing the voice at any time under circumstances connecting it with the alleged speaker” satisfies Rule 901’s authentication/identification requirement). At the evidentiary hearing, Goldsmith conceded that the tapes were “crucial” to the government’s case and that without them, the government would have had a “much more difficult time” proving that Drones was part of a conspiracy. Thus, going into trial, Goldsmith knew that the government had a key piece of direct evidence linking Drones to a conspiracy, and he knew that his client had to challenge the validity of that piece of evidence. The district court held that, under these circumstances, Goldsmith should have pursued both expert and layperson identification testimony in an effort to determine whether the voice on the tape was in fact Drones’s voice, and that his failure to do so constituted constitutionally deficient conduct.

We turn first to counsel’s decision not to investigate certain layperson identification testimony. Goldsmith testified that his decision not to play the tape for Johnson was based on his concerns that Johnson’s testimony (1) would have been impeachable due to his familial relationship with Drones and (2) could have opened the door to questioning about Drones’s prior narcotics convictions. Goldsmith chose not to play the tapes for Krisna because he believed that her testimony would have been cumulative given her alibi testimony. We find that these concerns were valid and his decisions were part of a plausible trial strategy. Goldsmith was correct that, if Johnson testified, he could have opened the door to questions about Drones’s prior drug convictions. Counsel’s decision not to pursue evidence that could be “double-edged in nature [was] objectively reasonable and therefore does not amount to deficient performance.” Lamb v. Johnson, 179 F.3d 352, 358 (5th Cir.1999) (quoting Rector v. Johnson, 120 F.3d 551, 564 (5th Cir.1997)); Kitchens v. Johnson, 190 F.3d 698, 702-03 (5th Cir.1999) (finding that counsel’s decision not to investigate mitigating evidence of child abuse, alcoholism, and mental illness was sound trial strategy). Similarly, Goldsmith determined that Krisna’s identification testimony would not have been of additional help to Drones’s defense given that she had already provided alibi testimony that, if believed, would have made it impossible for Drones to have made the recorded phone calls. This determination was not unreasonable under the circumstances. Since Goldsmith would not have presented either Krisna or Johnson’s testimony at trial, his decision not to investigate the substance of that potential testimony was reasonable.

We find Goldsmith’s failure to investigate the availability of expert spectro-graphic analysis of the tapes more troubling. Having ruled out two possible sources of identification testimony in Kris-na and Johnson, Goldsmith was still faced with a significant piece of damaging evidence at trial. Under these circum stances, Goldsmith’s decision simply to abandon any investigation of the tapes might not seem reasonable. This is not, however, a case in which counsel simply fails to articulate a justification for his limited investigation. Here, our analysis is complicated by the fact that Goldsmith testified that he chose not to investigate spectrographic analysis for strategic reasons. Specifically, Goldsmith testified that, at the start of the case, he made the “strategic decision” to “attack the government’s case” rather than to “assume the burden of proof ... and try to disprove that it was Mr. Drones’s voice on the tape.” Goldsmith further testified that his choice of trial strategy was influenced by the controversial nature of expert voice analysis. On appeal, the government argues, inter alia, that Goldsmith’s decision not to pursue spectrographic analysis was objectively reasonable given the questionable reliability of expert voice identification evidence.

We note as a preliminary matter that Goldsmith’s testimony raises serious doubts as to whether he actually knew enough about the relevant law to have made a reasoned legal decision based upon it. See Loyd v. Whitley, 977 F.2d 149 (5th Cir.1992) (finding deficient performance where “counsel’s decision had not been made after a thorough investigation of the law; [counsel] was unaware of the law”). Specifically, Goldsmith testified that that his decision was based in part on his belief that “at the time [of trial], the admissibility of this type of evidence was controversial based on the Fifth Circuit law that was in place at the time.” Our review of this circuit’s case law indicates that we have never before addressed the admissibility of spectrographic evidence.

The government takes Goldsmith’s argument a step further and argues that, “regardless of whether Goldsmith acted for the reasons he said he acted,” he was reasonable in failing to investigate expert voice identification testimony because the reliability of such evidence was questionable at the time of trial. We recognize that counsel is not required to pursue a line of defense that he knows will be fruitless. Cf. Nealy v. Cabana, 764 F.2d at 1178 (finding deficient performance where counsel “did not testify that such efforts would have been fruitless ... [but] simply failed to make the effort to investigate”). Here, however, there is virtually no indication in the record that Goldsmith ever did enough research into expert voice analysis to make a determination regarding the merits of a defense based on expert testimony. Rather, Goldsmith testified that he decided at the start of the case that he would not investigate any means by which he could affirmatively disprove that Drones’s voice was on the tapes. He then relies on this “strategic” decision to justify his failure to conduct even a preliminary investigation into the availability of expert witnesses in the Texas area. 6

Nonetheless, despite our reservations as to whether Goldsmith actually knew enough about the law surrounding the admissibility of spectrographic evidence to have made a reasoned legal decision based upon it, see Loyd, 977 F.2d at 158 (finding deficient performance where “counsel’s decision had not been made after a thorough investigation of the law; [counsel] was unaware of the law”), we find that Goldsmith’s failure to investigate can be constitutionally deficient only if it resulted in the exclusion of competent evidence. 7 Given the current state of the law regarding the admissibility of expert voice identification testimony and the expert testimony presented at the evidentiary hearing, we cannot say that counsel’s choice of strategy was unreasonable and therefore deficient.

At the time of Drones’s trial, four of our sister circuits had upheld the admissibility of spectrographic evidence, see United States v. Smith, 869 F.2d 348, 351 (7th Cir.1989) (“Under the Frye [v. United States, 54 App.D.C. 46, 293 F. 1013 (1923)7 test, several other circuits have held expert testimony concerning spectrographic voice identification admissible.... We join these circuits today, and hold that expert testimony concerning spectrographic voice analysis is admissible in cases where the proponent of this testimony has established a proper foundation.”); United States v. Williams, 583 F.2d 1194, 1201 (2d Cir.1978) (finding spectrographic analysis admissible under a modified version of the Frye test of admissibility); United States v. Baller, 519 F.2d 463, 465-67 (4th Cir.1975) (same); United States v. Franks, 511 F.2d 25, 32-34 (6th Cir.1975) (same), and one appellate court had refused to admit expert voice identification testimony, see United States v. Addison, 498 F.2d 741, 745 (D.C.Cir.1974); see generally United States v. McDaniel, 538 F.2d 408, 413 (D.C.Cir.1976) (finding that court was bound by prior ruling that voice identification evidence was inadmissible but noting trend towards admissibility of “voice-prints”). However, all of these cases were decided before Daubert v. Merrell Dow Pharmaceuticals, in which the Supreme Court set out the applicable standard for determining the admissibility of scientific evidence. See Daubert, 509 U.S. 579, 589-90, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1.993) (holding that Federal Rule of Evidence 702 superceded the “Frye standard” of admissibility of scientific evidence and that under Rule 702, the district court had to determine that proffered expert testimony was both reliable and relevant); id. at 593-94, 113 S.Ct. 2786, 125 L.Ed.2d 469 (setting out five non-exclusive factors to assist the district court in determining the reliability of proffered scientific evidence). No federal case decided postDaubert has considered the admissibility of expert voice identification testimony. 8

In addition to the fact that the state of the law concerning expert voice identification was ambiguous, the expert testimony presented at the evidentiary hearing demonstrates that spectrographic analysis is— and was at the time of Drones’s trial — of questionable scientific validity. Most notably, at the hearing, Koenig testified that there is no proven theoretical basis for the basic underlying premise that one person’s voice is truly unique and therefore identifiable. He further stated that this has resulted in a precipitous drop in the number of expert practitioners over the past few decades, from fifty to sixty practitioners in the 1970’s to roughly a dozen experts at the time of Drones’s trial. While Cain testified that expert voice identification testimony has been used extensively in state and federal courts over the past thirty years, he also testified that he did not know if spectrographic evidence was widely accepted by the relevant scientific community. He also acknowledged that numerous factors — including a defendant’s ability to disguise his own voice — could affect the reliability of expert analysis.

Thus, even assuming that expert voice identification would have met the Daubert standard, at trial, expert testimony that voice analysis would have indicated a “probable exclusion” of Drones as the unidentified voice on the tape would have almost certainly have been accompanied by, at the very least, a rigorous cross-examination of Drones’s expert. Most likely, the government would have also presented Koenig’s testimony that spectro-graphic analysis was, at best, a dwindling science. Beyond this, any expert testimony that Goldsmith could have presented would have been further impeached by the substantial circumstantial evidence of Drones’s participation in the drug conspiracy.

In sum, given the uncertainty of the current state of the law regarding the reliability and admissibility of expert voice identification evidence, and the vulnerability of such expert testimony at trial, we simply cannot say that Goldsmith’s choice of strategy — and specifically his decision to attack the government’s case rather than to rely on expert voice identification testimony — was unreasonable. 9 Accordingly, his representation of Drones was not constitutionally deficient.

Additionally, Drones cannot prevail under Strickland because he is unable to prove that he was prejudiced by his counsel’s actions. See Strickland, 466 U.S. at 694, 104 S.Ct. 2052, 80 L.Ed.2d 674. In order to satisfy the prejudice prong of Strickland, Drones must show “more than the mere possibility of a different outcome.” Ransom, 126 F.3d at 723. Rather, a petitioner must present “evidence of sufficient quality and force to raise a reasonable probability that,” had it been presented to the jury, the outcome would have been different. Id.; see also Strickland, 466 U.S. at 694, 104 S.Ct. 2052, 80 L.Ed.2d 674 (holding that in order to show prejudice, defendant must demonstrate that “there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different”). We find that, under the facts of this case, the district court erred in concluding that Drones met this substantial burden. More precisely, we find that given the existence of other circumstantial evidence against Drones, Drones cannot demonstrate that Goldsmith’s failure to investigate and present expert voice identification testimony prejudiced his defense. 10

While the government tapes were clearly important to proving the overall conspiracy, this is not a case where a determination of guilt rested entirely on this evidence. To the contrary, here, even assuming Drones’s voice was not on the tape, there was additional significant circumstantial evidence in the record to support the jury’s convictions.

In order to prove a drug conspiracy, the government must present evidence of (1) an agreement between two or more persons to violate narcotics laws, (2) a defendant’s knowledge of the agreement, and (3) a defendant’s voluntary participation in that agreement. See United States v. Chavez, 119 F.3d 342, 347 (5th Cir.1997). The agreement that forms the basis of a conspiracy charge can be implicit, and a jury may infer its existence from circumstantial evidence. See United States v. Thomas, 12 F.3d 1350 (5th Cir.1994); see also Chavez, 119 F.3d at 347 (“A defendant’s presence and association with other members of the conspiracy, when supported by other evidence, may be used to support a finding of conspiracy.”); United States v. Gallo, 927 F.2d 815, 820 (5th Cir.1991) (“The Government is not required to prove the existence of the conspiracy and the agreement between the co-conspirators and the defendant by direct evidence, but may present circumstantial evidence, such as the co-conspirator’s concerted actions, from which the jury can infer that a conspiracy existed.”). The essential elements of an aiding and abetting claim are (1) association with a criminal drug venture, (2) participation in the venture, and (3) action by the defendant that, in some way, tries to make the venture succeed. See Chavez, 119 F.3d at 347. Typically, the evidence that supports a conspiracy conviction supports an aiding and abetting conviction. See id. at 347.

Here, Officer Haire testified that he observed Drones sitting in the driver’s seat of the Mustang during the drug transaction. While Drones was in the car, the Freddies weighed eighteen pieces of crack cocaine. Haire further testified that at some point during the transaction, Drones stepped out of the Mustang and stood beside the car, and that when the officers approached the vehicle, Drones attempted to flee the parking lot. When the officers searched the vehicle, they found packages of crack in the front passenger compartment of the vehicle. Viewed as a whole, this evidence of concerted action and flight is strong support for the jury verdict. Cf. Gallo, 927 F.2d at 820 (“The existence of the agreement, the defendant’s participation in the conspiracy may be inferred from the ‘development and collocation of circumstances.’ ”); United States v. Thorn, 917 F.2d 170, 173 (5th Cir.1990) (“If Thorn understood the unlawful nature of the activity and knowingly or intentionally joined it on at least one occasion, that was sufficient to convict him of conspiracy, even though he played only a minor role in the scheme.”). The same evidence could also establish Drones’s association with a criminal narcotics venture and his participation in it. See United States v. Gallo, 927 F.2d 815, 822 (5th Cir.1991) (“The same evidence that establishes [a defendant’s] knowing and voluntary participation in the conspiracy also establishes his association with the criminal venture and participation in it.”). Specifically, the fact that eighteen pieces of crack cocaine were weighed while Drones sat in the driver’s seat of the car is strong circumstantial evidence of his association with and participation in the sale of that crack.

The fact that — even if Goldsmith had conducted a thorough investigation — at trial, the jury would have been presented with considerable circumstantial evidence of Drones’s guilt leads us to conclude that Drones was not prejudiced by counsel’s deficient conduct.

Ill

For the foregoing reasons, we REVERSE the judgment of the district court and REMAND with instructions to deny habeas relief.

Notes

. Arnold Freddie pled guilty to all three counts. Vernon Freddie pled guilty to the conspiracy and aiding and abetting charges and therefore was only tried for the firearm charge.

. The government presented two additional trial witnesses: (1) a chemist who testified that the substance found at the scene was crack cocaine; and (2) an ATF agent who testified that the gun found in the car was operational.

. Cain’s qualifications are extensive. At the time of trial, he was the president of Applied Forensic Technologies International. Prior to that, Cain spent twenty-two years working for various governmental agencies as a forensic specialist. Cain has been certified in voice identification and/or tape analysis by the Treasury Department, the International Association for Identification, and the American College of Forensic Examiners.

. Voice identification analysis consists of both "critical listening,” which essentially requires an expert to carefully listen to the qualities of two voices, and spectrographic analysis. In the latter phase, a computer-based instrument compares the frequencies and amplitudes of a voice on a questioned recording with those of a known sample. Both Cain and the government’s expert, Bruce Koenig, testified that there were seven possible results of a voice identification analysis: identification, probable identification, possible identification, inconclusive, possible elimination, probable elimination, and elimination.

. Koenig's credentials are also exemplary. At the time of trial, he was a private consultant involved in audio and video analysis. Prior to entering private practice, Koenig was employed as a supervisory special agent in the FBI where he also conducted audio and visual analysis.

. For example, Goldsmith testified that he did not make a motion for appointment of funds for an expert voice identification analysis because his “decision on the trial strategy made that motion unnecessary.” He also testified that, to his knowledge, there was no local expert witness but that he "made the strategic choice before needing to attempt to locate an expert.”

. While none of our previous cases explicitly hold that a failure to investigate can be deficient only if such failure leads to the exclusion of admissible evidence, the underlying assumption in our prior cases is that an investigation would have resulted in, at the very least, admissible evidence. See, e.g., Moore v. Johnson, 194 F.3d at 604 (finding failure to investigate defendant's background and the facts underlying an "accidental shooting” theory was professionally unreasonable); Bryant, 28 F.3d at 1418 (holding counsel’s failure to investigate alibi witness and eyewitnesses to the crime amounted to constitutionally deficient performance); Loyd, 977 F.2d at 157-59 (holding failure to obtain psychological analysis of defendant to present during sentencing phase of trial was not reasonable). In contrast, we are unable to locate any federal case in which a court held that an attorney rendered a constitutionally deficient performance due to his failure to investigate evidence that he would not have been able to present at trial.

. We note that the one state court to address the admissibility of spectrographic evidence post-Daubert determined that spectrographic analysis was admissible. See State v. Coon, 974 P.2d 386 (Alaska 1999).

. We do not reach the question of whether expert voice identification testimony is admissible under Daubert.

. Because we find that Goldsmith’s failure to investigate layperson testimony from Krisna and Johnson was reasonable, the government correctly isolates the issue here as whether Goldsmith’s failure to investigate voice identification testimony was prejudicial to Drones’s defense.