ORDER DENYING THE GOVERNMENT'S MOTION TO EXCLUDE THE TESTIMONY OF MICHAEL H.L. HECKER

KAY, District Judge.

INTRODUCTION

The government now moves to exclude the expert testimony of Michael H.L. Hecker, Ph.D. from trial. The defendant Ati So'o seeks to call Dr. Hecker on his behalf as an expert in the field of spectrographic voice identification. After hearing four days of testimony by Dr. Hecker and Special Agent Barbara Kolhus of the Federal Bureau of Investigation, and after reviewing the exhibits and the memoranda of law provided by the parties, the Court denies the motion and will allow experts from both sides to testify at trial.

BACKGROUND

In this prosecution for conspiracy and extortion under the Hobbes Act, 18 U.S.C. § 1951, an important piece of evidence for the government is a threatening tape recorded message of some 65 words left on the alleged victim's telephone answering machine. Although the caller did not identify himself, the prosecution alleges that the message was left by defendant So'o.

Mr. So'o denies making the phone call, and he has notified the government that he intends to call Dr. Hecker as an expert witness on this issue. If permitted to testify, Dr. Hecker will state that in his opinion, based upon his use of spectrographic voice analysis, the unknown caller is not Mr. So'o. It is uncontroverted that the identity of the person who left the threatening message is a key element of the prosecution's case against Mr. So'o and that the recorded message is the only direct evidence that the government has to support its allegation that Mr. So'o was involved in the alleged extortion scheme.

The method of voice identification by spectrographic analysis is based upon the principle that a person's voice can be effectively identified and distinguished when its characteristics are registered by an instrument known as a "spectrograph". A spectrograph produces a readout or linear representation of the voice on a "spectrogram".

A spectrogram is . . . a graphic display of sound in three dimensions: frequency in hertz, time in seconds, and different levels of energy. Spectograms display the fundamental frequency (pitch) of a voice and its resonating characteristics, as well as rate of speech. Different formations of words result in different patterns on the spectrogram. Spectograms are made by transferring a recorded voice to a machine, called a sound spectrograph, that transforms the sounds on tape into electrical coordinates. These coordinates manifest themselves as different shadings on paper.

The government's motion rests upon three separate grounds: 1) that the testimony is inadmissable because spectrographic analysis is a novel scientific technique which has not gained sufficient acceptance in its field; 2) that the evidence should be excluded because it will confuse and mislead the jury; and 3) that Dr. Hecker is not qualified to testify as an expert witness.

DISCUSSION

The controlling test for the admissibility of expert testimony on spectrographic voice identification is set forth in Frye v. United States, 293 F. 1013, 1014 (DC Cir. 1923) as follows:

"Just when a scientific principle or discovery crosses a line between experimental and demonstrable stages is difficult to define. Somewhere in this twilight zone the evidential forces of the principle must be recognized, and while courts will go a long way in admitting expert testimony deduced from a well-recognized scientific principle or discovery, the thing from which the deduction must be made must be sufficiently established to have gained general acceptance in a particular field in which it belongs."

In a recent case, the 7th Circuit noted that it continued to affirm and to apply the Frye test and that, under the standard, numerous other circuits have held expert testimony concerning spectrographic voice identification admissible. See, United States v. Williams, 583 F.2d 1194, 1198-1201 (2nd Cir. 1978), cert. denied, 439 U.S. 1117, 99 S.Ct. 1025, 59 L.Ed.2d 77 (1975); United States v. Baller, 519 F.2d 463, 465-467 (4th Cir.), cert. denied, 423 U.S. 1019, 96 S.Ct. 456, 46 L.Ed.2d 391 (1975); United States v. Franks, 511 F.2d 25, 32-34 (6th Cir.), cert. denied, 422 U.S. 1042, 95 S.Ct. 2654, 45 L.Ed.2d 693 (1975). The 7th Circuit decided it would join those other circuits in holding that expert testimony concerning spectrographic voice analysis is admissible in cases where the proponent of this testimony has established a proper foundation. The court pointed out that only one circuit has held that such testimony is inadmissible ( see United States v. Addison, 498 F.2d 741 (D.C. Cir. 1974)), and that even that Circuit has subsequently acknowledged that it might be time to reexamine Addison in light of the increased reliability and general acceptance of spectrograms. See United States v. McDaniel, 538 F.2d 408, 413 (D.C. Cir. 1976). It is notable that in Smith and in most of the circuit cases cited above, the government was the proponent of the expert testimony concerning spectrographic voice analysis.

As the government asserted in Smith, while the federal courts heavily favor admissibility, "state courts are more evenly split on the issue, although the majority favor the admission of voice spectrography." (Exhibit to Defendant's Memorandum at 41). Similarly, in Williams, supra, where the government was again the proponent of admissibility, the 2nd Circuit ruled that the weight of judicial authority supports the admissibility of spectrographic voice identification evidence. 583 F.2d at 1197. The Ninth Circuit has not directly addressed this issue of admissibility. However, there are three opinions in which the court has discussed voice spectrogram evidence, yet in none of them has the court held that such evidence was inadmissible per se. See United States v. Sukumolachan, 610 F.2d 685 (9th Cir. 1980); United States v. Goldstein, 532 F.2d 1305 (9th Cir.), cert. denied, 429 U.S. 960, 97 S.Ct. 384, 50 L.Ed.2d 327 (1976); and United States v. Turner, 528 F.2d 143 (9th Cir.), cert. denied, 423 U.S. 996, 96 S.Ct. 426, 46 L.Ed.2d 371 (1975). Furthermore, commentators favor admissibility. See, e.g., 5 J. Weinstein M. Berger, Evidence, § 901(b)(5)[01] (1976), at pp. 901-69 to 73; and McCormick, Evidence at 491 (1972).

The Ninth Circuit, in applying the Frye test, has ruled that the proponent of scientific evidence "that has yet to gain general judicial recognition `has the burden of laying a proper foundation showing the underlying scientific basis and reliability of the expert's testimony.'" United States v. Gillespie, 852 F.2d 475, 480 (9th Cir. 1988). In both Gillespie and United States v. Gwaltney, 790 F.2d 1378, 1382 (9th Cir. 1986), the Ninth Circuit reiterated that the trial court has wide discretion in determining whether particular scientific tests are reliable enough to permit expert testimony based upon their results.

This court finds first that pursuant to the Frye test, "the thing from which the deduction must be made [is] sufficiently established to have gained general acceptance in a particular field in which it belongs." As Dr. Hecker and Ms. Kolhus of the FBI have testified, spectrograms accurately reflect voice information, and a qualified examiner can identify individual voices. They disagree over how that information should be interpreted in this case, i.e., the examiner's deduction from such information. Indeed, the government argued in its brief in Smith that: "The basic scientific principle on which his testimony rests — that individual voices sound distinctive and the sound of a voice is reflected in its energy distribution which is recorded by the spectrogram — is uncontroverted. The area of controversy lies in the deduction made from that principle. . . ." Frye, of course, does not require such deductions to be "generally accepted." (See Exhibit to Def.'s Memo. at 39). The government further argued: "Thus, as long as the underlying principle is valid, an inferential deduction will be permitted even though no deduction can be absolutely reliable." ( Id. at 37). The government concluded in its Smith argument that "since the sound spectrograph does nothing more than plot these differences that are apparent to all, the `principle' on which it rests is no more speculative than the principle that permits lay witnesses to identify voices, despite intra speaker variation." ( Id. at 41).

This Court also concludes that the defendant has met his burden of establishing the scientific basis and reliability of the proposed expert testimony. See Gillespie, 852 F.2d at 480. The admissibility of spectrographic evidence under these circumstances has been affirmatively testified to in various cases by Dr. Hecker, Dr. Ladefoged, a linguist at UCLA, Dr. Nakasone, and Lt. Nash of the Michigan State Police. Admissibility in this case is also supported by the Tosi study and the Michigan State study which were discussed at length during the hearing. Additionally, the FBI has attested to a relatively low error rate, as evidenced by a 1986 study authored by the their expert Bruce Koenig. See "Spectographic Voice Identification: A Forensic Study", J. Acoust.Soc.Am. 79(6) at 2088.

Dr. Hecker defined the relevant scientific community to evaluate the validity and reliability of voice spectrography as comprised primarily of scientists in speech sciences, acoustics, and phonetics, and those who have familiarity and experience with voice spectrograms. The government offered no evidence on the relevant scientific community. The Smith court found the relevant scientific community included not only those who utilize spectrographic voice identification techniques, but also linguists, psychologists and engineers. 869 F.2d at 352. The Williams court was presented with a list of 10 scientists classified as favoring use of spectrographic analysis in the courtroom and 17 scientists opposed. The court pointed out that selection of the "relevant scientific community" appears to influence the result, with acceptance of the technique appearing strong among scientists who work with spectrograms and weak among those who had not, and with some of the scientists now classified as favoring previously having been opposed. 583 F.2d at 1198.

The 1979 study of the National Academy of Sciences ("NAS") made the following recommendation: "We recommend that if evidence on voice identification is admitted in court — and we take no position on admissibility — then the inherent limitations in the method and in the performance of examiners should be explained to the fact finder, whether the judge or the jury, in order to protect against overvaluation of the evidence." The government's brief in Smith, referring to the 1979 NAS study, asserted "that report is hardly compelling evidence, since its authors pointedly took no position as to whether spectrographic analysis was sufficiently reliable to warrant its admission into evidence." Exhibit to Def.'s Memo. at 39. The government went on to argue: "In any event, the committee's ultimate recommendation was that more study was needed in this area before definitive conclusions as to reliability could be reached. . . . That recommendation is now eight years old. As Dr. Nakasone pointed out at [the Smith] trial, the subsequent FBI study amply demonstrated the reliability of this technique." Id. at 40. The government also asserted that "the Tosi study has been generally accepted both in the literature and by the courts as authoritative." Id. at 35.

In proceeding with its analysis, the court in Smith found that other circuits have focused upon whether the technique is (a) reliable and (b) likely to mislead the jury. 869 F.2d at 351. Regarding the reliability of such evidence, the Court observed that other circuits have concluded "unanimity of opinion is not necessary among the scientific community to deem evidence reliable." Id. at 352. "`A determination of reliability cannot rest solely on a process of counting [scientific] noses.' The technique moreover need not be infallible to be reliable." Id. at 352, quoting Williams, 583 F.2d at 1183. As stated in Williams, "unanimity of opinion in the scientific community, on virtually any scientific question, is extremely rare. Only slightly less rare is a strong majority." 583 F.2d at 1198.

Citing to the Williams opinion, the Smith court noted several indicia of reliability, including potential rate of error, the existence and maintenance of standards, the care and concern with which a scientific technique has been employed, its analogous relationship with other types of scientific techniques, and characteristics the variability of which will lead to different, rather than similar, results. 869 F.2d at 352. In Williams, the court found that these indicia were present with regard to spectrograms. 583 F.2d at 1198-1200.

As stated, the Williams court found that one indicator of the evidential reliability is the potential rate of error. It cited that Dr. Tosi arrived at a false identification rate of 6.3%, which was reduced to 2.4% when doubtful comparisons were eliminated. Id. at 1198. Dr. Hecker relied upon the Tosi study in determining an estimate of his potential error rate, which he believes is 12% or less, given the safeguards that he has utilized. He compared the circumstances of this case to the situation defined by the Tosi study as "open trials — non-contemporary matching spectrograms — fixed context." See Def's Exhibit D-3 at 2038. The FBI's 1986 study of some 2,000 spectrograms showed a 0.53% error rate for false eliminations and an error rate of 0.31% for false identifications. Ms. Kolhus of the FBI testified that she was 80% to 90% confident in her conclusion, while her supervisor Mr. Koenig was over 90% confident in his conclusion, with both of them being of the opinion that defendant So'o was the unknown speaker. Dr. Tosi's study shows that the false identification error rate in "open trials — non-contemporary matching spectrograms — clue words spoken in fixed context" is 4.27% (Ex. D-3 at 2038). The study records that "each subpanel of examiners was forced to reach a common decision (identification or elimination) in each trial. . . ." ( Id. at 2036).

The Tosi study concludes that its findings suggest that if a trained non-professional examiner, using only visual inspection of spectrograms for purposes of identification and excluding any kind of listening, is forced to reach a positive decision in each case (devoting approximately 15 minutes to complete the task), the expected error of false identifications would be approximately 6% and the expected error of false eliminations would be approximately 13%. And, if the examiners had not been forced to produce a positive decision when uncertain, they would have reached a positive conclusion on only 74% of the cases, with approximately a 2% error of false identifications and approximately a 5% error of false eliminations. It is further hypothesized that if, in addition to visual comparisons of spectrograms, the examiners had been allowed to listen to the unknown voices, the errors might have been further reduced. Tosi felt that the field study performed at the Michigan Department of State Police reinforced that hypothesis. (Ex. D-3 at 2041). On the other hand and as acknowledged by Dr. Hecker, there might be a relatively equivalent counter balance of such a reduction of error rate by a failure to accurately replicate real world conditions.

The Smith court cautioned that "because of the apparent objectivity of opinions with a scientific basis, the jury may cloak such evidence in an `aura of mystic infallibility'". But the court concluded that the tendency of such evidence to mislead the jury is reduced by factors such as the comprehensibility of the technique, the ability of the jury to make the same comparisons as the expert, and the judges' instructing the jury to discredit the evidence if they find it unconvincing. Moreover, the evidence can be attacked by cross-examination and refutation. This court agrees with the conclusion reached by the Seventh Circuit in Smith: "A thorough examination of the record reveals both that this technique is not 100% infallible and that the entire scientific community does not support it . . . however, neither infallibility nor unanimity is a precondition for general acceptance of scientific evidence under Frye." 869 F.2d at 354.

The Smith court further stated: "We also note, as the government points out in its brief, that spectrographic identification is similar to lay identification of voices, which is admissible in this circuit." Id.

In its Smith brief, the government showed that spectrographic voice identification is a good deal more reliable than polygraphy. The government pointed out that in a polygraph examination, indices such as blood pressure and perspiration are charted on the assumption that they permit the examiner to infer a quality — credibility — which cannot itself be directly measured; whereas in spectrographic analysis, the speaker's pitch and other aural qualities are directly measured by the spectrogram. The government concluded that accordingly, the "spectrographic analysis attempts to measure a far more determinate quality in a far more determinable way than does polygraphy." (Exhibit to Def's Memo. at 39). In Williams, the court similarly distinguished the reliability of voice spectrography from that of polygraphy, since "in spectrography the examiner merely compares spectrograms reflecting the purely physical characteristics of a voice. In polygraph analysis, the examiners must go on to extrapolate a judgment of something not directly measured by the machine, i.e., the credibility of the person examined." 583 F.2d at 1199 n. 9. The court also observed that spectrography is qualitatively different from polygraph evidence and the "skill of a polygraph examiner, the kinds of questions asked, natural variations and blood pressure among individuals, and in how accustomed they are to lying, are unpredictable variables that make the polygraph technique far more speculative then a spectrographic analysis." Id. The court felt that spectrogram evidence would not mislead or confuse the jury since "the critical step in spectrographic voice analysis is a simple step of visual pattern-matching, a step easily comprehended and evaluated by a jury." Id. at 1199.

In this case, Dr. Hecker testified that the scientific community looks for the following assurances before it will accept a scientific technique: (1) validity, that the technique measures what it purports to measure; and (2) reliability, that the same result will be achieved if the experiment is performed again (replication). Dr. Hecker testified that scientists did not question the results of Tosi's laboratory results, but some scientists were concerned whether "real world conditions" could be replicated accurately. Dr. Hecker agreed that the spectrographic technique must employ controls and safeguards in each case in order to fall within the charted area of the conditions of the laboratory test.

Dr. Ladefoged agrees with Dr. Hecker that aural spectrographic analysis is accurate if there are safeguards. ( See, Govt's Ex. 47 at 13, summarizing 1985 testimony in People v. Siervonti). Dr. Ladefoged proposes the following safeguards: (1) two plus minutes of each speech sample; (2) a signal to noise ratio where the signal is higher by 20 decibels; (3) a frequency of 3,000 hertz or better; (4) an exemplar in the same words, the same rate, in the same way, spoken naturally and fluently; and (5) a responsible examiner. Dr. Ladefoged believes there is general acceptance given his safeguards, and "he believes there is now more agreement." Id. The court finds that Dr. Hecker has met all of these safeguards. While Dr. Ladefoged referred to two minutes of speech sample, Dr. Hecker explained that the subject sample of 35 seconds is more than adequate since it is a compact sample without pauses or interruptions in speech. Counsel for the government concurs that the speech sample is adequate time-wise, and the FBI itself requires only 22 seconds of speech sample. As far as background noise, Dr. Hecker testified there was very little background noise and that the tape of the unknown voice was of unusually good quality. Ms. Kolhus of the FBI concurred in this opinion.

Dr. Hecker listed the following conditions and safeguards present in the instant case: (1) the recording of the unknown voice was of good quality, (2) the exemplars of defendant So'o were made to the same recording machine, (3) several words were repeated in the unknown voice message, giving good intra-speaker variability analysis, (4) two exemplars were made of defendant So'o several months apart, thereby providing Dr. Hecker with the means to test whether defendant So'o had attempted to disguise his voice as well as further analysis of intra speaker variability, (5) Dr. Hecker felt that it was possible through his cooperative speaker to achieve the same identical rhythm, intonation, and emotional environment, (6) he concluded that the emotional stress in the unknown voice tape was moderate, and (7) Dr. Hecker was able to employ his thirty-some years of training and experience in analyzing the spectrograms and tapes, utilizing both global pattern and speech features approaches. The speech feature approach relates to the anatomical features and habitual use patterns, which are hard to manipulate. Since the FBI has concluded from its spectrographic analysis that Mr. So'o was the unknown caller, it evidently did not find that the emotional content in the unknown call had any impact on the FBI's ability to make an accurate examination. Dr. Hecker and Ms. Kolhus of the FBI both testified that there is generally less intra-speaker variability than there is inter-speaker variability.

Both Dr. Hecker and the FBI utilized the aural spectrographic method, listening to the tapes as well as viewing the spectrograms. Dr. Hecker testified that spectrographic analysis plus comparing the tapes by listening is more accurate than either method by itself. He pointed out that the spectrogram provides a more permanent picture to study, and an examiner may be mislead more easily by a disguised voice in listening. The Michigan Police Department study and the FBI utilized aural comparison in addition to spectrographic analysis, which was also recommended by the NAS.

Second, in the alternative, as noted above there has been general judicial recognition of the admissibility of spectrographic evidence to identify voices. See, Smith 869 F.2d at 351, Williams 583 F.2d at 1197, Exhibit to Defendant's Memorandum at 41. As the court stated in Gillespie, "we apply the Frye test to expert opinion testimony that is `based on a novel scientific technique,'" as distinguished from scientific evidence that has "gain[ed] general judicial recognition." 852 F.2d at 480. Thus it would appear that inasmuch as the admissibility of spectrographic evidence to identify voices has received general judicial recognition, it is no longer considered novel within the Frye test and consequently the test is inapplicable.

Finally, the Ninth Circuit has set forth the general test regarding the admissibility of expert testimony as being whether the jury can receive "appreciable help" from such testimony. United States v. Gwaltney, supra, 790 F.2d at 1381. In its Smith brief, the government also states that the federal rules of evidence are helpful in explaining how the Frye test should be applied.

The rules define "relevant evidence" as "evidence having any tendency to make the existence of any fact that is of consequence to the determination of matter more probable or less probable than it would be without the evidence." Federal Rules of Evidence 401. The rules then state that relevant evidence should be admitted unless its probative value "is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury, or by considerations of undue delay, waste of time, or needless presentation of cumulative evidence. Federal Rules of Evidence 403. Finally, Rule 702 provides: "if scientific, technical or specialized knowledge will assist the trier of fact to understand the evidence or determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, training or education, may testify thereto in the form of an opinion or otherwise." Federal Rules of Evidence 702. Thus the Frye test must be construed as requiring sufficient acceptance of the principle so that the evidence can truly be considered of assistance to the jury: "expert testimony is admissible only when the specialized knowledge of an expert will assist the trier of fact in understanding the evidence for determining a fact in issue." (Def's Ex. at pages 36 and 37).

In this case, the testimony of Dr. Hecker and the government's expert on spectrographic analysis will be of "appreciable help" to the jury. As argued by the defense, the message on the alleged victim's phone machine is the only direct evidence of defendant Ati So'o's involvement in the extortion scheme. In addition, as Dr. Hecker testified the reliability of aural identification alone is subject to question. In this case where both sides have qualified experts to testify as to their opinions of the identity of the speaker in question, the jury will be aided by such testimony. As stated in U.S. v. Addison, 498 F.2d at 744, "[T]he Frye test protects prosecution and defense alike by assuring that a minimal reserve of experts exist who can critically examine the validity of a scientific determination in a particular case."

In view of the foregoing, this court will deny the government's motion to exclude Dr. Hecker's testimony under Rule 403 of the Fed.R. of Evidence, and allow spectrographic analysis for voice identification into evidence in this case.

Finally, based upon the testimony and voir dire examination at the hearing, the Court finds that both Dr. Hecker and Ms. Kolhus of the F.B.I. are qualified to testify as experts on spectrographic voice identification.

CONCLUSION

For all the reasons stated above, the Court denies the government's motion to exclude the expert testimony of Dr. Hecker from evidence at trial. Accordingly, the Court will permit each side to present testimony on the subject of spectrographic voice identification.