OPINION
EASTAUGH, Justice.
I. INTRODUCTION
A jury found George Coon guilty of making three terroristic telephone calls. We must decide whether the superior court erred in admitting opinion evidence, based on a voice spectrographic analysis, that Coon made those calls. Applying the Alaska Rules of Evidence and the standard for admitting novel scientific opinion evidence, articulated by the United States Supreme Court in Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993), we hold that it was not error to admit that evidence, and affirm Coon’s conviction.
II. FACTS AND PROCEEDINGS
Three messages threatening David Rudolph’s life were left on Deborah Rudolph’s telephone answering machine. Deborah, Coon’s former daughter-in-law, recognized the recorded voice as Coon’s. The State charged Coon under former AS 11.56.810(a)(2) with terroristic threatening.
The State retained a voice analysis expert, Steve Cain, who compared the voice on the answering machine with verbatim voice exemplars provided by Coon. At Coon’s 1992 trial, the superior court held a hearing outside the jury’s presence to determine whether Cain’s testimony would be admissible under Frye v. United States, 293 F. 1013 (D.C.Cir.1923). The superior court found that Cain’s testimony was relevant and would assist the jury. It therefore held that the evidence was admissible under Alaska Evidence Rule 702. The court also concluded that spectrographic analysis of voices satisfied the Frye test for admitting novel scientific evidence, and that the identification of voices by analyzing spectrograms had been generally accepted by courts and was a reliable technique. The jury then heard Cain express his opinions that Coon made the first telephone call, and that there was a high probability Coon also made the second and third calls. Cain described for the jury the scientific foundation for his opinions. The jury found Coon guilty of terroristic threatening.
On appeal, the Alaska Court of Appeals held that the appellate record did not support admitting Cain’s testimony at trial.1 It reasoned that the State had not presented evidence concerning the relevant scientific community and whether that community generally accepted voice spectrographic analysis. It noted that the superior court had made no findings on those topics, contrary to the requirements of Frye. It remanded, reasoning that the State might be able to establish that voice spectrographic analysis met the Frye [389]*389standard for admissibility. Citing Contreras v. State, 718 P.2d 129 (Alaska 1986), where we adhered to the Frye standard, the court of appeals declined to address the State’s argument that the Frye standard was no longer viable in Alaska following adoption of the Alaska Rules of Evidence and announcement of the Daubert opinion.
Petitioning for hearing, the State asked us to consider the same arguments. It also argued that the trial court’s ruling satisfied Daubert. We granted the State’s petition. Following briefing and oral argument, we remanded with directions that the superior court enter findings of fact and conclusions of law about whether Cain’s testimony was admissible under Evidence Rule 703. The order also required findings of fact and conclusions of law about the admissibility of the proffered testimony under the Frye and Daubert standards. We retained jurisdiction. Entering separate findings and conclusions for each standard, the superior court ruled on remand that the evidence was admissible under both tests.
We then asked the parties and possible amici curiae to file briefs discussing whether to retain Frye or to adopt the test articulated in Daubert to determine the admissibility of scientific evidence. Five amici curiae — the Alaska Academy of Trial Lawyers (AATL), the Alaska Public Defender Agency (APDA), the Defense Counsel of Alaska (DCA), the Product Liability Advisory Counsel (PLAC), and Rex Lamont Butler and Associates (Butler) — submitted briefs thoroughly discussing the issues presented. We thank them for their valuable assistance.
III. DISCUSSION
A. Frye, Daubert, and the Rules of Evidence
The State asks us to abandon the scientific evidence test discussed in Frye and adopt the “more flexible” standard announced in Daubert. Whether to adopt a new standard for admitting evidence involves the interpretation of the Alaska Rules of Evidence and is therefore a legal question to which this court applies its independent judgment, adopting the rule most persuasive “in light of reason, precedent and policy.”2
Before Daubert was issued, Frye was the dominant standard for the admissibility of scientific evidence in federal and many state courts.3 'We adopted the Frye standard in Pulakis v. State, 476 P.2d 474, 478-79 (Alaska 1970).
The Frye court held the results of a crude lie detector test, based on the examinee’s blood pressure, to be inadmissible to show a criminal defendant’s innocence.4 The court’s entire analysis consisted of one paragraph and cited no supporting authority:
Just when a scientific principle or discovery crosses the line between the experimental and demonstrable stages is difficult to define. Somewhere in this twilight zone the evidential force of the principle must be recognized, and while courts will go a long way in admitting expert testimony deduced from well recognized scientific principle or discovery, the thing from which the deduction is made must be sufficiently established to have gained general acceptance in the particular field in which itbelongsl[5]
The court concluded that the systolic blood pressure deception test had “not yet gained such standing and scientific recognition among physiological and psychological authorities” as to be admissible into evidence.6
Seventy years later in Daubert, the United States Supreme Court ruled that the Federal [390]*390Rules of Evidence, adopted in 1975, had superseded the Frye “general acceptance” test for admitting scientific evidence.7 The plaintiffs in Daubert were minor children with serious birth defects.8 They alleged that the birth defects resulted from their mothers’ ingestion of Bendectin, a prescription anti-nausea drug, during pregnancy. They sought to introduce expert testimony that Bendectin was a teratogen, a substance capable of causing malformations in fetuses.9 The trial court refused to admit the testimony and entered a defense judgment.10 The Ninth Circuit affirmed, but the Supreme Court vacated the judgment and remanded.11 So ruling, the Court stated that “a rigid ‘general acceptance’ requirement would be at odds with the ‘liberal thrust’ of the Federal [Evidence] Rules and their ‘general approach of relaxing the traditional barriers to ‘opinion’ testimony.’ ”12 The Court concluded that the “austere” Frye “general acceptance” standard, “absent from, and incompatible with, the Federal Rules of Evidence, should not be applied in federal trials.”13
The Court delineated a standard that requires the trial judge to “determine at the outset, pursuant to [Federal Rule of Evidence] 104(a), whether the expert is proposing to testify to (1) scientific knowledge that (2) will assist the trier of fact to understand or determine a fact in issue.”14 This two-step inquiry requires a “preliminary assessment of whether the reasoning or methodology underlying the testimony is scientifically valid and of whether that reasoning or methodology properly can be applied to the facts in issue.”15 The Court also provided a non-exhaustive list of factors courts may use in making this inquiry. These included (1) whether the proffered scientific theory or technique can be (and has been) empirically tested (i.e., whether the scientific method is falsifiable and refutable); (2) whether the theory or technique has been subject to peer review and publication (although publication “is not a sine qua non of admissibility”); (3) whether the known or potential error rate of the theory or technique is acceptable, and whether the existence and maintenance of standards controls the technique’s operation; and, echoing Frye, (4) whether the theory or technique has attained general acceptance.16
The Court offered “general observations” concerning publication, peer review, and other factors, and how they pertain to scientific validity.
The inquiry envisioned by Rule 702 is, we emphasize, a flexible one. Its overarching subject is the scientific validity — and thus the evidentiary relevance and reliability— of the principles that underlie a proposed submission. The focus, of course, must be solely on principles and methodology, not on the conclusions that they generate.[17]
Daubert thus requires trial courts to ensure that scientific evidence is both relevant and reliable. The opinion is widely regarded as imposing a more rigorous “gatekeeper” function on trial courts than Frye did.18
In 1986, before the Supreme Court announced Daubert, but after we adopted rules of evidence in 1979 modeled after the Federal Rules of Evidence, we rejected an argument by the State in Contreras that the Federal Rules of Evidence had replaced [391]*391Frye.19 Recognizing that the federal rules did not mention the Frye test, we said that “[w]e believe it unlikely that this silence was meant to overturn long-established rules of admissibility based on Frye .20 Although we did not expressly say so, we implicitly held that the Frye standard was not inconsistent with the Alaska Rules of Evidence. We also noted that “we have held other pre-existing rules to survive adoption of the rules of evidence.”21 Noting that we had adopted Frye in Pulakis, we declined to question Frye.22
We have not revisited the question since deciding Contreras. We noted the existence of Daubert in Mattox v. State, Department of Revenue, 875 P.2d 763, 764 n. 2 (Alaska 1994), but declined to consider the issue. The Alaska Court of Appeals has commented on the conflict between Frye and Daubert, but in conformity with Contreras, it has continued to apply the Frye standard.23 In its Memorandum Opinion and Judgment in this case, the court of appeals stated that, given our adherence in Contreras to the Frye standard, “we believe that ... the proper course of action is for us to follow the existing standard.”24
We now reconsider the standard in light of Daubert’s holding that Frye is inconsistent with the Federal Rules of Evidence. Because the trial court found Cain’s opinion evidence to be admissible under both tests, and assuming that it applied each test correctly, it can be argued that there is no reason for us to resolve a legal issue that is not dispositive of the outcome of this petition. We nonetheless choose to reach the Frye /Daubert issue because the parties and the amici have thoroughly and skillfully discussed the reasons why we should or should not adopt Daubert in place of Frye. Postponing a decision for another day would be unlikely to give us the benefit of better advocacy. Further, to the extent that this issue relates to our rules of evidence, it implicates our rule-making responsibility.
Although we are not bound by the Supreme Court’s conclusion, its analysis of the corresponding federal rules is helpful and, moreover, has triggered a wealth of useful critical comment.25
B. Ex Post Facto Prohibition
Before deciding whether Frye remains ascendant, we consider, and reject, Coon’s argument that the federal and state constitutional prohibitions on ex post facto legislation apply to judicial decisions thus barring the application of the 1993 Daubert decision to his 1992 trial.26 We reach this conclusion because there is no support for Coon’s proposition in federal or state law.
On its face, the federal ex post facto prohibition applies only to legislative acts, not judicial decisions.27 We construe [392]*392our state prohibition no differently than the federal prohibition.28 Unforeseeable judicial enlargements of criminal statutes have been struck down as violative of due process when applied retroactively.29 But that is not the case here.
Even if we were to apply the prohibition on ex post facto laws to judicial decisions, changes to rules governing the admissibility of evidence do not violate it. In Thompson v. Missouri, 171 U.S. 380, 387, 18 S.Ct. 922, 43 L.Ed. 204 (1898), the United States Supreme Court stated that
[W]e cannot perceive any ground upon which to hold a statute to be ex post facto which does nothing more than admit evidence of a particular kind in a criminal case upon an issue of fact which was not admissible under the rules of evidence as enforced by judicial decisions at the time the offense was committed.[30]
The prohibition on ex post facto laws has been construed as applicable only to penal legislation.31
Coon also asserts that retrospective application of Daubert to his case would deny him substantive and procedural due process of law and violate his right to equal protection. His cursory discussion of these issues is inadequate to preserve them. We consider them waived.32
C. The Alaska Rules of Evidence
The State argues that we should abandon the Frye standard. It asserts that Frye has become outdated and inadequate for modern litigation, where many cases involve sophisticated scientific data and knowledge. It argues that Frye uses social, rather than scientific, criteria for determining reliability and validity when reviewing a novel scientific technique. This causes trial courts simply to “count hands” to determine whether scientists in the relevant scientific community accept the technique as reliable, and “abdicates” judicial responsibility for determining admissibility to scientists uneducated in the law.
The State also argues that a few dissenters within a scientific community may prevent a finding of general acceptance, leading to over-representation of the dissenters’ views. In addition, the State contends that Frye ⅛ conservative nature causes a “gross time lag” between the development of a new scientific technique and its judicial admissibility. This can cause certain cutting edge science to become obsolete before it is admissible under Frye.
Our new evidence rules became effective in 1979, nine years after we decided Pulakis.33 We there approvingly quoted the Frye standard in holding that polygraph test results had been properly excluded.34
Several of our evidence rules bear on the admissibility of scientific evidence. Evi[393]*393dence Rule 104(a) assigns to the trial court the duty to determine preliminary questions concerning the qualification of a person to be a witness and the admissibility of evidence. Evidence Rule 401 defines what evidence is relevant. Evidence Rule 403 allows exclusion of relevant evidence for such reasons as prejudice, confusion, and waste of time. Evidence Rule 702 allows experts to offer helpful opinion testimony.35 Evidence Rule 703 allows experts to base opinions on facts or data of a type reasonably relied upon by experts in the field.36 Thus, expert opinion evidence is admissible if the trial court (exercising its authority under Rule 104(a)) determines that (1) the evidence is relevant (Rule 401); (2) the witness is qualified as an expert (Rule 702(a)); (3) the trier of fact will be assisted (Rule 702(a)); (4) the facts or data on which the opinion is based are of a type reasonably relied upon by experts in the particular field in forming opinions upon the subject (Rule 703); and (5) the probative value of the evidence is not outweighed by its prejudicial effect (Rule 403).37
Nothing in our evidence rules requires or implies that any single criterion, let alone Frye ⅛ general acceptance standard, controls admission of scientific opinion evidence. The commentary, which predated Daubert, observed that Evidence Rule 703 “attempts to chart a path between the rigid approach of [Frye] and the minimal relevance approach of Rule 401.”38 It also mentioned several Daubert-like factors and noted:
Even though Rule 403 might be deemed sufficient protection against the dangers of relatively untested evidence, Rule 703 is drafted so as to remind trial judges that innovative attempts to offer expert evidence may involve evidence that is superficially attractive, but which is problematic for one or more of the following reasons: ... 3) while the expert evidence is plainly relevant, the rate of error associated with the technique that produced the evidence is unknown and the trier of fact is therefore unable to properly evaluate the evidence; 4) the expert evidence is the subject of great controversy among the nation’s experts and it would be inappropriate for a court or jury to resolve the controversy in any particular ease. See, e.g., People v. Kelly, 17 Cal.3d 24, 130 Cal.Rptr. 144, 549 P.2d 1240 (1976) (rejecting voiceprint evidence)[39]
Although the United States Supreme Court stated in Daubert that Federal Evidence Rule 702 is the “locus” for determining the admissibility of scientific evidence,40 the commentary to the Alaska Rules of Evidence provides support for the State’s view that Alaska Rule of Evidence 703 is also a source for an approach broader than the Frye standard.
Our evidence rules give trial courts both the authority and the responsibility to determine the admissibility of such evidence without being limited to the general acceptance standard. They preclude this inquiry from focusing exclusively on general acceptance or any other single factor. Our evidence rules contemplate a broader inquiry, allowing a proponent to establish admissibility even if general acceptance is absent, and allowing an opponent to challenge admissibility even if general acceptance is present.
Frye is potentially capricious because it excludes scientifically reliable evidence which is not yet generally accepted, and admits scientifically unreliable evidence which al[394]*394though generally accepted, cannot meet rigorous scientific scrutiny. Because the Frye test potentially excludes evidence that should be admitted under our rules, and also potentially admits evidence that should be excluded under our rules, we conclude that it is both unduly restrictive and unduly permissive. Just as the Supreme Court concluded in Daubert that the Frye test is inconsistent with the federal evidence rules,41 we conclude that it is inconsistent with the Alaska Rules of Evidence.
We reach this conclusion despite our prior approval of the Frye standard. “[T]he judicial doctrine of stare decisis accords the prior holdings of the highest courts of this State precedential value while still permitting the reconsideration of legal issues when conditions warrant.” 42
We have stated that we will overrule a prior decision only when we are “clearly convinced that the rule was originally erroneous or is no longer sound because of changed conditions, and that more good than harm would result from a departure from precedent.” 43 It is our view that these requirements are met with respect to our decision in Contreras.
First, we are clearly convinced that we erroneously decided Contreras for the following reasons. Contreras assumes that the Federal Rules of Evidence did not change the Frye test. Daubert has subsequently shown that assumption to be incorrect. Further, our decision in Contreras ignored Alaska Evidence Rule 703, which employs a “reasonably relied upon by experts” standard in contrast to Frye ⅛ “general acceptance” standard. This oversight in Contreras is all the more surprising because the commentary to the Rule 703 makes clear an intent to promulgate a rule different from Frye: “The rule attempts to chart a path between the rigid approach of [Frye] and the minimal relevance approach of Rule 401.”
Second, we conclude that the “more good than harm” requirement is also met. As noted above, Frye may exclude scientifically reliable evidence while admitting unreliable evidence. It is desirable to replace Frye with a rule not suffering from these deficiencies. Any harm as may be done by overruling Contreras is short term and limited. It concerns only cases in which the Frye /Dau-bert controversy has been raised that are pending in the trial courts or on direct review at the time of this decision.44 In those cases, if reliance on Frye is found to be other than harmless error, a new trial may be required. But a new trial may not be necessary, because on remand the trial court may determine that the questioned evidence meets the standard set out in this opinion.
The limitations of the general acceptance standard have been extensively catalogued.45 Assuming that some of the strictures of Frye were ameliorated in some courts pre-D<m-bert,46 we nonetheless conclude that it is better to resolve admissibility disputes by referring to our modern evidentiary foundation, the Alaska Rules of Evidence, than by trying [395]*395to salvage or remold a “rigid” standard that is fundamentally inconsistent with our rules,
How should Alaska trial courts assess the reliability and relevance of proffered scientific evidence? The factors identified in Dau-bert provide a useful approach: (1) whether the proffered scientific theory or technique can be (and has been) empirically tested (i.e., whether the scientific method is falsifiable and refutable); (2) whether the theory or technique has been subject to peer review and publication; (3) whether the known or potential error rate of the theory or technique is acceptable, and whether the existence and maintenance of standards controls the technique’s operation; and (4) whether the theory or technique has attained general acceptance.47
Other factors may apply in a given ease. After the Supreme Court issued its decision in Daubert, the Ninth Circuit suggested two ways to satisfy Daubert ⅛ requirement that the testimony be “derived by the scientific method [or] ... based on scientifically valid principles.”48 As described by Kesan, “either (a) the expert’s proffered testimony must grow out of prelitigation research, or (b) the expert’s research must be subjected to peer review.” 49 Kesan, giving the example of “independent” research funded by tobacco companies, appropriately notes the danger of a hidden litigation motive.50 Nonetheless, publication is at least more likely to provoke scrutiny and response, and reveal methodological deficiencies.
Alaska Evidence Rule 702 is similar to New Mexico’s equivalent rule.51 In 1993 the New Mexico Supreme Court adopted the Daubert standard in State v. Alberico, 116 N.M. 156, 861 P.2d 192, 203-04 (1993). Other state supreme courts with similar evidence rules have also adopted the Daubert standard.52
But other states, whose evidence rules also mirror the federal rules, have rejected Dau-bert and retained Frye.53 California and New York courts have also retained a Frye-type standard; they have no rule similar to Alaska Evidence Rule 702.54 Those decisions do not convince us that we should retain Frye.
In invoking our evidence rules, adopting Daubert, and limiting Frye, we reject arguments and dire predictions supporting the status quo.
Burden on trial judges. We first reject concerns that Daubert will make the trial courts’ gatekeeping role unduly burdensome. This concern is founded on a perception that Daubert requires judges to determine the reliability of the expert’s scientific methods, whereas the general acceptance standard allows courts to defer to the judgment of scientists.55
The notion that scientists are better suited than judges for assessing scientific reliability may appear initially persuasive, given that [396]*396trial judges are rarely trained in science and given concerns that they are already overburdened. Closer consideration reveals that the notion is misleading and irrelevant. First, scientific reliability is not necessarily congruent with judicial reliability. This is illustrated by Frye itself. The “general acceptance” standard does not define scientific reliability; it is simply a judicial construction. And for reasons noted above, it is a flawed judicial construction. Second, the rules of evidence must be applied by trial judges, subject to review for abuse of discretion. It is for the trial court to determine whether the expert is qualified to testify and the proffered evidence is admissible.56 Determining reliability for judicial purposes is unavoidably the responsibility of trial courts, and should not be delegated to an expert’s peers.
The burden, especially when considering novel scientific evidence, may well be substantial. We nonetheless conclude that it is one the trial courts (and appellate courts on review) must bear. But we are not convinced that the burden will be as onerous as some predict. We suspect that most difficult disputes will be limited to evidence that is rationally disputable; we expect that relatively little effort will be required to determine the admissibility of most scientific evidence, because most will be either patently reliable or unreliable.
As for the difficult disputes, the courts may reduce the judicial burden and increase the accuracy of their admissibility decisions by selecting independent expert witnesses.57 They can also appoint expert advisors.58 Such advisors can provide valuable guidance to courts determining reliability of proffered scientific evidence. Commentators have identified various concerns about using technical advisors.59 Their suggestions — how to select an advisor, how to define the expert’s duties, and how the expert should act — are worthy of consideration.60
“Junk science.” Several amici argue that juror susceptibility to the persuasive power of scientific evidence mandates a conservative reliability standard, such as Frye ⅛ general acceptance test, to prevent admission of “junk science.”61
In State v. Carter, 246 Neb. 953, 524 N.W.2d 763, 777-78 (1994), the Nebraska Supreme Court stated that the Frye rule was intended to ensure the reliability of scientific evidence because: (1) lay jurors can be overly impressed by science; (2) lay jurors lack the capacity to evaluate scientific evidence critically; and (3) lay jurors are likely to give “junk science” more weight than it deserves. The court, citing New Mexico and Arizona decisions, recognized “the complex nature of [397]*397DNA evidence and the need to protect against unproven and potentially erroneous and misleading evidence,” and declined to adopt the Daubert standard for the admissibility of DNA evidence.62
The Frye test is arguably a safeguard against evidence based upon specious scientific techniques, especially in criminal trials where the defendant’s right to a fair trial is crucial.63 Nevertheless, the Frye standard has also been criticized for being easily manipulated by courts when deciding whether or not to admit certain evidence. “The lack of a definitional framework for ‘field’ and ‘general acceptance’ allowed courts seeking to admit scientific evidence to confine the ‘field’ of pertinent inquiry narrowly to a specialty within a broader scientific discipline in order to demonstrate ‘general acceptance.’”64 For example, in Commonwealth v. Dykus, 367 Mass. 191, 327 N.E.2d 671, 675-78 (1975), the record contained evidence of a dispute about the acceptance of voice spectrography in the scientific community; the court resolved the dispute by limiting the applicable scientific community to “those who would be expected to be familiar with its use.”65
We are not convinced that “junk science” is more likely to be admitted under Daubert than under Frye. Tost-Daubert reported decisions suggest that courts are acting with restraint, and are giving rigorous consideration to the reliability of scientific evidence.66 Furthermore, Frye also potentially permits admission of unreliable scientific evidence, because a methodology that has been generally accepted might nonetheless have been discredited during a Daubert inquiry.
We also suspect that junk science poses less risk to fair trials than a more common aspect of admitting scientific evidence. Even when they apply identical methodologies that satisfy both Daubert and Frye, dueling experts can reach conflicting, irreconcilable conclusions. Indeed, Daubert has been criticized because it assumes that reliability should be determined by examining the methodology, without regard to the truth of the scientific conclusions.67 Even when they apply the same methodology, experts for opposing parties may reach different results. Likewise, well-qualified experts may make rational conflicting choices in deciding which reliable methodology to apply. Juries, relatively ill-trained to resolve such conflicts on subtle or complex scientific grounds, may resort to credibility assessments which are unduly simplistic. Opinions of a practiced forensic expert may prevail over those of a true academic who makes a less impressive appearance on the witness stand. Assuming proper execution of the judicial gatekeeping function, we think the risk to fair trials posed by junk science is lower than the risk posed by jurors’ difficulty in critically assessing scientific evidence that is based on reliable methodologies.
Relitigation of evidence admissible under Frye. Butler’s amicus brief asserts that since the Daubert standard applies to all scientific knowledge, and is not limited to “novel” scientific evidence, evidence deemed admissible under Frye may now be found inadmissible under Daubert. Butler argues that this will lead to increased litigation over the admissibility of scientific evidence, and to a case-by-case determination of admissibility, with the possibility of inconsistent or unpredictable decisions. Butler suggests that Daubert may [398]*398affect forensic sciences, such as fingerprint, handwriting, and hair comparison analyses, that are now admissible under Frye. While Butler frames this as a negative result of adopting Daubert, the New Mexico Supreme Court noted:
Contrary to the assertion ... that the Frye test places the responsibility of determining scientific validity upon scientists, in practice too many courts reference reported case law to determine what is generally accepted in the scientific community. It is improper to look for scientific acceptance only from reported case law because that amounts to finding a consensus in the legal community based on scientific evidence that is sometimes many years old[68]
Further, as the State notes, when an area of expertise is well-known and has been fully considered by the courts, a trial court may take judicial notice of its admissibility. The Supreme Court advocated this approach.69 Moreover, general acceptance remains a factor under Daubert.70 It also seems unlikely that methodologies that were admitted under Frye and that remain generally accepted in the appropriate community will be excluded, absent affirmative evidence of unreliability.
D. The Admissibility of Voice Spectrograph Evidence under Daubert
On remand, the trial court found that Cain’s testimony regarding the applied forensic technique of using spectrography to analyze and identify voices was admissible under both Frye and Daubert. Coon and the APDA contend that the trial court erred in finding Cain’s testimony regarding voice spectrography admissible under either test. Given our adoption of Daubert in Part III.C, there is no reason to review admissibility under Frye.
1. Standard of review
We review a trial court’s ruling on the admissibility of expert testimony for abuse of discretion.71
Although the parties agree on this standard of review, two amici curiae, Butler and the APDA, argue that we should review de novo a trial court’s decision to admit or exclude scientific evidence. So arguing, APDA relies on a passage from Pulakis, where we stated that “[o]n the basis of our study of the judicial authority and academic literature in this area, we conclude that the results of polygraph examinations should not be received in evidence over objection.”72
In addition, both Butler and APDA cite Contreras, in which the court of appeals stated that it was “free to exercise [its] independent judgment” when reviewing the trial court’s decision to admit evidence recovered from an eyewitness during hypnosis because the findings regarding memory and the effects of interrogative techniques on eyewitness testimony and hypnosis in general are legislative facts, rather than adjudicative facts.73
Kesan and the dissenting opinion propose adopting a hybrid standard of review, reviewing de novo a trial court’s findings on the scientific knowledge prong of the Daubert test and retaining an abuse of discretion [399]*399standard for the second prong relating to relevance or “fit” of the facts of the particular case to the scientific technique employed.74 They assert that implementing de novo review of scientific validity allows for more uniform adjudication at the trial and appellate levels, and for development of distinct validation criteria for expert testimony relating to different scientific or technical disciplines.75
We see no present reason to adopt a new or hybrid standard for reviewing rulings on the admissibility of scientific evidence. Abuse of discretion is the standard applicable to other evidentiary rulings.76 Such rulings are best left to the discretion of the trial court. A determination of reliability under Daubert is no different.
The dissent reaches a different conclusion because it begins with the premise that the scientific validity of a technique is a legal issue which does not turn on case-sensitive facts.77 This premise does not adequately take account of the reality of the judicial process and the variable state of science. The New Mexico Supreme Court rejected the same argument advanced here by the dissent. In rejecting the de novo standard of review that court noted:
This reasoning assumes, however, that the record on appeal contains all of the relevant, most recent data concerning the scientific method, and that assumes too much. It also assumes that there is always a reservoir of scientific literature that an appellate court might independently reference in a de novo review. The abstract validity of a scientific technique should not vary from court to court, but how the proof of such validity is communicated will often vary from presentation to presentation. Some experts are more skillful and more well-informed than others just as some lawyers are more skillful and more well-prepared than others. In addition, ■ the state of science is not constant; it progresses daily. For example, what might have been true about [post traumatic stress disorder] in the early 1980s when it was first addressed in published opinions might not have been true in 1992.... [78]
We recognize that different trial judges, in exercising their discretion, may reach different conclusions about scientific reliability. But we disagree with the dissent’s suggestion that the inconsistency will be of such magnitude as to “compromise the integrity of the judiciary in the eyes of the public.”79 Despite such predictions in the academic litera-tee, the majority of the federal circuits have chosen to apply the abuse of discretion standard when reviewing district court decisions under Daubert.80 In addition, the Supreme Court recently approved the abuse of discretion standard in General Electric, Co. v. Joiner, 522 U.S. 136, 118 S.Ct. 512, 517, 139 L.Ed.2d 508 (1997).
The principal reason for adopting the Dau-bert standard is to give the courts greater flexibility in determining the admissibility of expert testimony, so as to keep pace with science as it evolves. We think the abuse of discretion standard of review best comports with these aims, and we choose to apply it here.
[400]*4002. Admissibility of Cain’s voice spectro-graphic analysis under Daubert
In making its findings on remand, the trial court discussed each factor the Supreme Court articulated in Daubert. The trial court found that the technique of spectrographic voice identification had been empirically tested “on numerous occasions by many scientists during approximately the past twenty years.” The court determined that voice spectrography had been subjected to peer review and publication, but noted that the technique’s reliability was debated in the scientific literature. It also found that when properly performed under the “stringent standards” of the International Association for Identification by a qualified, trained scientist or technician, voice spectrography has a known error rate of less than one percent. The trial court found that because voice spectrography has been subjected to empirical testing, it is both falsifiable and refutable, and that testing has not refuted the technique or shown that it is falsified. The court determined that when voice spectrography is properly performed by a qualified person, it has attained widespread acceptance within the relevant scientific community — amongst forensic scientists and scientists in acoustics and speech-related fields with experience using the technique.
The trial court also found that the reasoning and methodology underlying Cain’s testimony were scientifically valid, and that Cain had properly performed the voice spectro-graphic analysis in this case. The trial court stated that Cain’s testimony was relevant in that it assisted the jury in determining a key issue at trial, namely the identification of the person who left the threatening messages on Deborah Rudolph’s answering machine. The trial court therefore concluded that Cain’s testimony was sufficiently relevant and reliable to be admitted under Daubert.
The trial court made these findings on remand after reviewing the record, including Cain’s testimony.
Because this is our first review of a trial court’s application of Daubert, we discuss Cain’s evidence at some length. Cain testified at the 1992 hearing about his qualifications as a voice spectrograph expert. He has a bachelor of science degree in engineering and two master’s degrees in forensic science; he had completed two years of work toward a doctoral degree in criminology. He had continued his post-graduate education by attending seminars at the state and national levels dealing with tape examination and other aspects of forensic evidence analysis. He had given talks and seminars throughout the United States on voice identification and tape analysis and had published “a number of articles” concerning the legal aspects and results of research studies on the reliability of voice spectrograph analysis.
He owned and operated Applied Forensic Technologies International, Inc., a forensic laboratory that analyzes physical evidence. Before opening his private laboratory, Cain worked for ten years as a document and voice print examiner for the Secret Service in its Washington, D.C., crime laboratory, and for three years for the Internal Revenue Service as a supervisor at its Illinois Crime Lab, where he started a voice identification program.
He was certified by the International Association for Identification (IAI) as a voice identification specialist and he served on the certification board of that organization. Based on this testimony offered outside the jury’s presence, the trial court found that Cain was qualified as an expert under Alaska Evidence Rule 702 in the field of “applied forensic techniques of voice prints.”
Cain then explained that in performing a voice analysis, he conducts both an aural and a spectrographic analysis of a voice. Cain discussed factors which might affect the results of voice analysis, the potential error rate of voice analysis, and the reliability of the technique. He described the standards adopted by the IAI in 1992, and stated that he adhered to these standards, and that he sends his work to another examiner for an independent review of the accuracy of his findings. Cain estimated that approximately thirty-one to thirty-four states have accepted voice spectrograms, commonly known as voiceprints, as a positive means of forensic identification. Cain testified that voice spec-trographic analysis has been subjected to [401]*401empirical testing from the time of its inception over twenty years prior. Cain testified about various scientific studies that concluded that voice spectrographic analysis was a valid and reliable forensic identification technique.
The scientific literature submitted to us by amicus APDA would also support a finding that spectrographic analysis has been empirically tested. We conclude that the trial court did not err in finding on remand that this technique has been subjected to empirical testing.
Cain conceded that several studies question the validity and reliability of the technique. Although many of the studies questioned the reliability of the technique, the trial court did not err in finding on remand that the technique had been subjected to peer review and publication, since a general consensus is not a mandatory requirement under Daubert.
Cain also testified that research studies have calculated the known error rate for voice spectrographic analysis to be less than one percent when the technique is performed properly by a scientist skilled in the technique.81 Cain stated that in order to make the sample as accurate as possible by eliminating “intraspeak variation” (the internal factors that affect a person’s voice, such as the time of day, the amount a person has eaten, or the person’s mood), the scientist conducting the analysis should take numerous exemplars and attempt to duplicate the conditions of the original recording. The trial court did not err in finding on remand that the known error rate for voice spectro-graphic analysis performed properly by a skilled scientist in the field was sufficiently low to make this evidence reliable.
General acceptance remains a factor under Daubert. The trial court on remand found that “[t]he technique of voice identification using spectrography when properly performed by a qualified person has attained widespread acceptance within a relevant scientific community — forensic scientists and scientists in acoustics and speech related fields who had experience using the technique.”
In determining the admissibility of voice spectrograph evidence under the general acceptance test, several courts have defined the composition of the “relevant scientific community.” In State v. Gortarez, 141 Ariz. 254, 686 P.2d 1224, 1233 (1984), the Arizona Supreme Court defined the relevant scientific community as “disinterested and impartial experts in many fields, possibly including acoustical engineering, acoustics, communications electronics, linguistics, phonetics, physics, and speech communications.” The Gor-tarez court noted that this list was merely suggestive and not all-inclusive.82
Neither party questions the trial court’s determination of the relevant scientific community in this case. Based upon the findings of other courts as to the relevant scientific community for the technique of forensic voice identification using spectrography, we conclude that the trial court did not abuse its discretion in determining the relevant scientific community.
Coon and the APDA argue that voice spectrograph analysis is not generally accepted within the relevant scientific community, and has been ruled inadmissible by numerous courts applying Frye. The APDA has submitted several articles questioning the reliability of voice spectrograph analysis when used as anything other than an investigative tool. The APDA also included a copy of an Alaska superior court decision in State v. Gomez, No. 3AN-S88-5190 Cr. (March 23, 1990), in which Superior Court Judge Joan M. Katz found that voice spectrographic identification evidence was inadmissible because it was not [402]*402generally accepted within the relevant scientific community.
In determining whether evidence is generally accepted within the scientific community, courts have generally looked to three sources for guidance: (a) judicial opinions; (b) scientific literature; and (c) expert testimony presented at an evidentiary hearing.83
Under Frye many courts have determined that voice spectrographic evidence is not admissible.84 But other courts have allowed voice spectrographic evidence under standards other than Frye.85 And several courts purporting to apply Frye have also allowed such evidence.86 Overall, it is inconclusive whether there is a judicial consensus that voice spectrographic evidence is generally accepted within the relevant scientific community.
The scientific literature cited by the APDA permits a conclusion that there is significant disagreement among experts in the field of voice spectrographic analysis regarding the reliability of the technique. As the State notes, no scientific literature was submitted to the trial court for review, but Cain testified about several articles and studies addressing voice spectrographic analysis, and conceded that the reliability of the technique was disputed among members of the relevant scientific community.
Cain cited, among others, a study conducted in 1986 by Bruce E. Koenig of the Federal Bureau of Investigation. Koenig there noted that in a survey of 2,000 voice identification comparisons made by FBI examiners under actual forensic conditions, meaningful decisions were made only 34.8% of the time, with a known error rate of 0.31% for false identifications and 0.53% for false eliminations, with an overall error rate of 0.43%.87 In a 1993 article, Koenig noted that problems still persist in the spectrographic voice identification field, such as “separate sets of certified examiners making high confidence decisions for both identification and elimination in the same case.”88 Koenig also stated that even with the establishment of new standards by the IAI in 1992, the use of spectrographic voice technique is on the decline, and “[t]he most important legal difference is the FBI’s policy not to provide testimony on spectro-graphic comparisons due to the inconclusive nature of the examination and the unknown error rate under specific investigative conditions.” 89
Although it is not clear that voice spectro-graphic analysis has attained general acceptance within the relevant scientific community, we do not find that the trial court clearly erred in making its general acceptance finding, or abused its discretion in ruling that the evidence satisfied Daubert. Consequently, we conclude that the trial court did not err in finding the voice spectrographic evidence admissible under Daubert. It therefore did not abuse its discretion in admitting this evidence at Coon’s trial.
IV. CONCLUSION
For these reasons, we overrule Contreras, and hold that the Alaska Rules of Evidence supersede the Frye test.
We adopt the Daubert standard for determining the admissibility of scientific evidence. We hold that the voice spectrograph analysis evidence was admissible under Dau-bert and the Alaska Rules of Evidence in this [403]*403case. We therefore AFFIRM Coon’s judgment and conviction.
BRYNER, Justice, not participating.