United States v. Williams, 443 F. Supp. 269 (S.D.N.Y. 1977)

Robert B. Fiske, Jr., U. S. Atty., S. D. N. Y., New York City, for plaintiff.

Richard Rosenbaum, New York City, for defendant Williams.

Murray Mogel, Federal Defender Services, New York City, for defendant Manning.

VINCENT L. BRODERICK, District Judge.

On June 3, 1977, the government filed an indictment charging defendants Williams and Manning with violations of the federal narcotics laws. Specifically, defendants were charged with a conspiracy to violate the federal narcotics laws in violation of 21 U.S.C. § 846 (1972), and with possession and distribution of heroin, in violation of 21 U.S.C. § 841 (1972). A superseding indictment, charging defendants with the same offenses, was filed November 30, 1977.

In support of its case against defendant Williams, the government announced its intention to offer in evidence a tape recording of a telephone conversation allegedly between defendant Williams and a police officer; a voice exemplar furnished by defendant Williams; testimony of an expert witness who has compared the voices on the original tape and the exemplar; and spectrograms used by the expert in making his comparison.

Defendant Williams opposed the admissibility of this evidence on the ground that sound spectrography is not a reliable scientific method.

A pre-trial hearing was held to determine the admissibility of the proffered evidence, at the conclusion of which I ruled that the expert testimony and the spectrograms would be admitted. This memorandum sets forth the factual and legal bases for that determination.

The standard for admission of a scientific method is whether the technique "[enjoys] general acceptance in the particular field in which it belongs." Frye v. United States, 54 App.D.C. 46, 293 F. 1013 (1923).

It is the judge's responsibility to initially assess whether there is a sufficient degree of general scientific acceptability to warrant admission of the evidence. United States v. Stifel, 433 F.2d 431, 438 (6th Cir. 1970), cert. denied, 401 U.S. 994, 91 S. Ct. 1232, 28 L. Ed. 2d 531 (1971). Once a judge concludes that expert testimony is to be permitted, the jury is entitled to review the evidence and accord it such weight as the jury feels is warranted. Feguer v. United States, 302 F.2d 214, 242 (8th Cir.), cert. denied, 371 U.S. 872, 83 S. Ct. 123, 9 L. Ed. 2d 110 (1962).

United States v. Alexander, 526 F.2d 161, 163-64 (8th Cir. 1975).

The standard for admission of expert testimony enunciated in Frye v. United States, supra, has been applied in cases involving the admissibility of spectrographic voice identification.

The Court of Appeals for the District of Columbia was presented with the issue of the admissibility of spectrographic voice identification as evidence in a criminal case in 1974. United States v. Addison, 162 U.S. App.D.C. 199, 498 F.2d 741 (1974). After reviewing the testimony presented to the District Court, the Court of Appeals reversed the finding of the lower court and held the evidence inadmissible on the ground that general acceptance of the scientific community with respect to spectrographic identification had not yet reached the level called for by Frye:

[W]hile portions of the record suggest that spectrogram analysis may become a useful tool for the resolution of questions of criminal liability, it is equally clear that techniques of speaker identification by spectrogram comparison have not attained the general acceptance of the scientific community to the degree required in this jurisdiction by Frye. Whatever its promise may be for the future, voiceprint identification is not now sufficiently accepted by the scientific community as a whole to form a basis for a jury's determination of guilt or innocence.

Id. at 745.

In 1975 the admissibility of spectrographic voice identification was sustained by the Courts of Appeals for the Fourth and Sixth Circuits.

In United States v. Franks, 511 F.2d 25 (6th Cir.), cert. denied, 422 U.S. 1042, 95 S. Ct. 2656, 45 L. Ed. 2d 693 (1975), the Court acknowledged the decision in Addison, supra but found that "the trend favors the *271 admissibility . . .." Id. at 33. The Court suggested that "those opposing the admissibility . . . can direct their criticisms toward the weight of such evidence." Id. In Franks, the defense did not produce a witness in opposition to the admissibility of the evidence. Defendant Williams has produced a witness in opposition. However, as discussed below, I find the arguments of the proponents of the use of this testimony persuasive, and certainly within the Frye standard.

The Court of Appeals for the Fourth Circuit, in sustaining the admissibility of spectrographic voice identification in United States v. Baller, 519 F.2d 463 (4th Cir.), cert. denied, 423 U.S. 1019, 96 S. Ct. 456, 46 L. Ed. 2d 391 (1975), outlined safeguards against the danger that the jury might accord undue weight to the scientific testimony. The suggested safeguards include evidence of probative value despite doubts within the scientific community about its absolute accuracy, competent witnesses available to expose the limitations of the identification technique, availability of tapes for aural comparison by the jury, and an instruction to the jury that they are to weigh the evidence presented. Id. at 466-67. See also United States v. Jenkins, 525 F.2d 819, 827 (6th Cir. 1975). As discussed below, all of these safeguards have been or will be utilized in this case.

In United States v. McDaniel, 176 U.S. App.D.C. 60, 538 F.2d 408 (1976), the Court of Appeals for the District of Columbia again refused to admit evidence of spectrographic voice identification, but with considerably less certainty than in Addison, supra. The Court recognized the trend favoring admissibility, and contemplated that it may be time to reexamine the holding of Addison "in light of the apparently increased reliability and general acceptance in the scientific community." Id. at 413. However, the Court was constrained to follow its prior holding in the absence of a clear showing of dramatic change in the opinion of the scientific community or en banc reconsideration of Addison. Id.

This court is not constrained by a prior decision holding spectrographic voice identification evidence inadmissible. The Court of Appeals for the Second Circuit has not ruled on this issue, although evidence of spectrographic voice identification has been admitted by courts in this district on at least two prior occasions. United States v. Carlos Valle, No. 75 Cr. 857 (S.D.N.Y.1975), United States v. Sisca, 361 F. Supp. 735 (S.D.N.Y.1973).

While some courts have been reluctant to admit testimony of spectrographic voice identification, commentators favor its admission. See Weinstein's Evidence, ¶ 901(b) (5) (01) at p. 901-69 (1976): "At this stage of development, given an adequate expert's testimony, admissibility is warranted."

McCormick goes further. He would not use the standard of general acceptance in the scientific community, but would admit any relevant conclusions which are supported by a qualified expert unless there are other reasons for exclusion, such as prejudice, misleading the jury, or undue consumption of time. McCormick on Evidence, ¶ 203 at 491 (2d Ed. 1972).

Based on the testimony discussed below, I find that there is a sufficient showing of acceptance by the scientific community of the use of spectrographic analysis for voice identification to warrant its admissibility. Further, I find that there will not be prejudice, or undue consumption of time. The jury will be given instructions appropriate to guard against any danger that it may be misled in its interpretation and utilization of the testimony.

At the hearing, both the government and defendant Williams presented expert testimony on the issues of sound spectrography in general and the specific spectrographic analysis performed in this case.

The government's primary expert witness offered in support of the reliability of spectrographic analysis of the human voice was Dr. Oscar Tosi. Dr. Tosi, who has previously testified in this district and elsewhere on the subject, was accepted, for the purpose *272 of the hearing, as an expert in the field of sound spectrography. Dr. Tosi explained the use of the spectrograph and the various studies he has conducted on sound spectrography in general and voice identification in particular.

A sound spectrograph consists essentially of a) a rotating drum onto which a section of graph paper is mounted, and b) an electric stylus that marks this paper as the drum rotates. The stylus is activated by the segment of sound that is being analyzed. The pattern that the stylus leaves on the paper records the duration, frequency and amplitude of the sound, the duration being measured along the horizontal axis of the graph, frequency along the vertical axis, and amplitude by the darkness of the lines that the stylus leaves.

The spectrograph thus produces a graphic or pictorial representation of three characteristics of a segment of sound its duration, frequency and amplitude. This representation is called a sound spectrogram.

The theory behind the use of the spectrograph in voice identification is that when a human speaker speaks a particular word or phrase, he or she produces a sound that has a certain duration, and a certain set of variations in frequency and amplitude. When the same speaker speaks the same word again, he or she will produce a sound with the same or similar characteristics, that is, a sound of about the same duration, with about the same variations in frequency and amplitude. The spectrograms of the two sounds will also be similar, since the spectrograms are merely graphic representations of these same sound characteristics. A visual comparison of the spectrograms of two sounds can therefore furnish a basis to determine whether the two sounds are made by the same speaker.

Initially there were many errors of false identification. Recently the errors in a forensic situation were 6.3%. However, Dr. Tosi indicated that the percentage of error could be reduced to approximately 2% if the examiner made an aural examination of the tapes as well as a visual examination of the spectrogram. Dr. Tosi readily admits that the method is not infallible, but a showing of infallibility is not required for a finding of admissibility; a showing of general acceptance in the particular field will suffice. United States v. Alexander, supra, 526 F.2d at 163, n.3.

The next government witness was Dr. Henry Truby, also accepted as an expert for the purposes of the hearing. Dr. Truby is a specialist in the field of acoustic phonetics, which is concerned with the instrumental analysis of speech sound. One instrument used is the sound spectrograph. Dr. Truby testified that if the reported signal is of a substantial fidelity and if the operator knows what he is doing, has had sufficient experience, has a sample adequate to his particular task, and is given sufficient time to make his study, voice identification through comparison of sound spectrograms is reliable.

He also testified that he had inspected the spectrograms made by the government's expert in this case, and that those spectrograms were of exceptionally high quality.

Dr. Tosi and Dr. Truby both rated highly the professional qualifications of Mr. Frederick Lundgren, who conducted the spectrographic analysis which the government seeks to introduce in this case. Mr. Lundgren is a voice print specialist employed by the U. S. Bureau of Alcohol, Tobacco and Firearms, and was the final witness who testified at the hearing on behalf of the government. Mr. Lundgren, accepted as an expert for the purposes of the hearing, had testified in court eight times prior to his appearance before me.

He compared the voice sample which the government had from the tape recording of a telephone conversation with an "unknown" person the questioned voice with a voice exemplar, made by tape recording a telephone call of defendant Williams. For the exemplar, defendant Williams repeated the language used by the questioned voice. Mr. Lundgren made spectrograms of the most natural parts of the samples. He compared them visually and also made an aural comparison of the tapes.

*273 There are five possible conclusions Mr. Lundgren could have arrived at: 1) a positive affirmative identification; 2) a probable affirmative identification; 3) inability to identify; 4) a probable lack of identification; and 5) a positive lack of identification. He testified that ten positive matches of specific spectrographic elements are regarded as the minimum for a positive affirmative identification; in this case, he found thirteen positive matches. Mr. Lundgren's conclusion was that there was a positive affirmative identification. In other words, he concluded that the "unknown" voice heard on the tape recording of the original telephone call and the voice on the exemplar, that of defendant Williams, are the voice of the same person. This positive affirmative identification was also based on his aural examination of the tapes.

On cross-examination Mr. Lundgren was questioned as to whether the spectrogram similarities could have resulted from the fact that the same words were used on the exemplar as on the questioned recordings. Mr. Lundgren testified that he found characteristics present which were attributable to the speaker and not to the word, such as similar formant angle and starting position.

Mr. Lundgren also testified that the fact that the tapes were from telephone conversations did not present a danger of misidentification. The use of the telephone does, he conceded, narrow the band of information obtained by the spectrograph. But he asserted that while this might lead to a failure to identify, it could not result in a false identification.

Defendant Williams' expert witness was Dr. Louis J. Gerstman, a professor of psychology, and of speech and hearing. Dr. Gerstman was accepted as an expert for the purposes of this hearing. Dr. Gerstman has worked with sound spectrograms since 1949, and has testified in various state courts as well as the federal court in this district. Dr. Gerstman is opposed to the use of spectrographic voice identification in court proceedings because he believes that it is possible for spectrograms of the same word by two different talkers to be closer visually than spectrograms of the same word spoken twice by the same person. He feels that further experiments on this issue are necessary. However, Dr. Gerstman acknowledged the clarity of the spectrograms used by Mr. Lundgren and had no questions about Mr. Lundgren's integrity.

Dr. Gerstman is also concerned that there is a mystique of scientific precision which may mask the ultimate subjectivity of spectrographic analysis. Dr. Gerstman's concerns are valid ones. However, they may be alleviated by action other than suppression of the evidence. The jurors are presented with the evidence as an aid to their determination of whether the voices are or are not the same, not as a substitute for that determination. The defendant may submit evidence questioning the reliability of the spectrographic identifications, and the jury will be instructed that it may consider all such evidence in determining what weight, if any, to give the evidence. United States v. Alexander, supra, 526 F.2d at 164.

I find that voice identification by aural comparison and spectrographic analysis has probative value; that the technique of spectrographic analysis has been accepted by a substantial section of the scientific community concerned; that the government's proposed expert in this field, Mr. Frederick Lundgren, is qualified; and that the jury will not be misled by such evidence. I have therefore ruled that evidence of spectrographic voice analysis and identification will be admitted in this case.

SO ORDERED.