Docket No. 1516.
June 25, 1974.
Appeal from Superior Court of Fresno County, No. 26331, Blaine E. Pettitt, Judge.
Allen Ruby, under appointment by the Court of Appeal, for Defendant and Appellant.
Evelle J. Younger, Attorney General, Edward A. Hinz, Jr., and Jack R. Winkler, Chief Assistant Attorneys General, William E. James, Assistant Attorney General, Joel Carey and Anthony L. Dicce, Deputy Attorneys General, for Plaintiff and Respondent.
Appellant, E.D. Law, was convicted by a jury of one count of making a false bomb report to a police officer in violation of Penal Code section 148.1, subdivision (a), a felony (count one), and of two counts of making telephone calls with intent to annoy in violation of Penal Code section 653m, subdivision (a), misdemeanors (counts three and four).
The conviction of the false bomb report charge was based primarily upon the admission into evidence, over objection, of spectrograms (voiceprints) and expert testimony pertaining to the spectrograms. This evidence, viewed in a light most favorable to the prosecution, positively identified appellant as the unknown caller who made the bomb threat.
Though appellant's estranged wife and his 14-year-old son also identified the voice of the bomb threat caller as that of appellant, the Attorney General conceded at oral argument that if we held the voiceprint evidence to be inadmissible, the trial court's admission thereof would necessarily be prejudicial error.
The misdemeanor convictions were grounded on nonvoiceprint evidence and will be treated separately in this opinion.
THE BOMB THREAT CONVICTIONIn 1968, in People v. King, 266 Cal.App.2d 437 [ 72 Cal.Rptr. 478], the court (Second Dist., Div. Two) held voiceprint evidence inadmissible. Between 1968 and December 1970, under the directorship of Dr. Oscar Tosi and financed by a $300,000 federal grant, further studies were conducted at Michigan State University on the voiceprint identification method. Based upon these studies and the foundation testimony of Dr. Tosi, Hodo v. Superior Court (1973) 30 Cal.App.3d 778 [ 106 Cal.Rptr. 547] (Fourth Dist., Div. Two) held voiceprint evidence admissible.
However, neither the King nor Hodo cases involved an effort to disguise or mimic voices. In the instant case, the bomb threat was received by Officer Holland of the Fresno Police Department from a man who stated that he had "placed a bomb in the Federal Building and Ed Law was next." Exemplars of appellant's voice were subsequently obtained both with and without his knowledge. On one occasion appellant volunteered to make a recording of the words uttered by the bomb threat caller. On that occasion, appellant was directed to and did mimic the voice and accent of the unknown caller. Appellant's estranged wife, who identified the bomb threat caller as appellant, testified that appellant was "very good at imitating many people, many voices," and that the police recording of the telephoned bomb threat was not appellant's "normal voice," "He is imitating someone. I would say, I would say he was trying to imitate a colored person. But, he's got a little Oakie in there too."
These recordings, together with a recording of the actual bomb threat, were sent to Lieutenant Ernest Nash of the Michigan State Police for spectrographic analysis. After laying the foundation for Lieutenant Nash's testimony by calling Dr. Oscar Tosi and Dr. Peter Ladefoged, the prosecution called Lieutenant Nash who testified that based upon a spectrographic and aural analysis of the tapes, the voice of the unknown caller who phoned in the recorded bomb threat to the police department was the voice of appellant "and it could be the voice of no other person."
Lieutenant Nash is a detective with the Michigan State Police in charge of the Voice Identification Unit of the Scientific Crime Laboratory. He holds no degrees but is a junior at Michigan State University, majoring in audiology and speech sciences. He has published several papers on voice identification and has given a number of talks at seminars, having analyzed more than 1,500 voices and 80,000 voiceprints and having testified in court as an expert a number of times. Dr. Tosi described him as "the best examiner . . . in the world. . . ."
Dr. Peter Ladefoged is a professor of phonetics at University of California at Los Angeles. He has a Master of Arts and a Doctor of Philosophy in phonetics from the University of Edinburgh in Scotland. He has published a textbook on acoustic phonetics and has also published numerous articles on experiments dealing with a spectrograph and related acoustical analysis of speech.
(1) The test for determining the admissibility of testimony based on a newly developed scientific experiment or principle was stated in Huntingdon v. Crowley (1966) 64 Cal.2d 647, 653-654 [ 51 Cal.Rptr. 254, 414 P.2d 382]: "`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 a 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 it belongs.' (Italics added.)"
Under this rule, courts have rejected various types of scientific tests, including (1) Kell-Cellano blood grouping tests ( Huntingdon v. Crowley (1966) 64 Cal.2d 647, 656 [ 51 Cal.Rptr. 254, 414 P.2d 382]; (2) expert testimony of a defendant's statements and actions while under hypnosis ( People v. Busch (1961) 56 Cal.2d 868, 878 [ 16 Cal.Rptr. 898, 366 P.2d 314]); (3) statements made under sodium pentothal, or truth serum ( People v. Jones (1959) 52 Cal.2d 636, 653 [ 343 P.2d 577], cert. den. 361 U.S. 926 [4 L.Ed.2d 350, 80 S.Ct. 364]; People v. Johnson (1973) 32 Cal.App.3d 988, 1001 [ 109 Cal.Rptr. 118]; (4) behavioral manifestations by persons with chromosomal abnormalities ( People v. Tanner (1970) 13 Cal.App.3d 596, 601 [ 91 Cal.Rptr. 656, 42 A.L.R.3d 1408]; and (5) polygraph tests ( People v. Carter (1957) 48 Cal.2d 737, 752 [ 312 P.2d 665]; People v. Ferguson (1969) 1 Cal.App.3d 68, 76 [ 81 Cal.Rptr. 418].)
In reviewing the trial court's application of this test, we recognize that a certain amount of deference should be paid to the trial judge's decision. The actual extent of appellate review of such matters, however, has received minimal analysis in the cases touching on the point. Unquestionably, the trial court enjoys great latitude in determining the qualification of an expert, and its determination will not normally be disturbed on appeal. ( People v. Busch (1961) 56 Cal.2d 868, 878 [ 16 Cal.Rptr. 898, 366 P.2d 314]; Huffman v. Lindquist (1951) 37 Cal.2d 465, 476 [ 234 P.2d 34, 29 A.L.R.2d 485]; Witkin, Cal. Evidence (2d ed. 1966) § 1175, p. 1088.) (2) While there have been some statements that the same amount of discretion rests in the trial court to determine whether a new scientific test or process has passed from the experimental into the demonstrable stage and has received general acceptance by recognized experts in the scientific field in which it belongs ( Hodo v. Superior Court, supra, 30 Cal.App.3d 778, 784-785), there is a view that the latter issue is one of law and that the courts should not subsume the question of qualifying the process of spectrographic voice identification under the question of qualifying the expert. (Comment, Evidence: Admissibility of Spectrographic Voice Identification, 56 Minn.L.Rev. 1235, 1245.) In any event, it appears settled that the court can and should take judicial notice of the case law and comments on the particular field of scientific endeavor and of articles from reliable sources that appear in scientific journals and other publications which suggest the possibility of error in tests and experiments. Such judicial notice is taken for the purpose of determining if the procedure has passed from the experimental to the demonstrable stage and has received general acceptance by recognized experts in the field. ( State v. Dantonio (1955) 18 N.J. 570 [ 115 A.2d 35, 39, 49 A.L.R.2d 460]; see 29 Am.Jur.2d, Evidence, §§ 103-107, pp. 134-137; McBaine, Cal. Evidence Manual (2d ed. 1960) § 504, p. 159.)
See also Huntingdon v. Crowley, supra, 64 Cal.2d 647, 654, 656, where it appears the court considered a broad spectrum of scientific literature on the subject being discussed.
Our appellate courts have in the past and should in the future quite properly show deep concern that the trier of fact might give an undue amount of credence to a new scientific development and be overwhelmed by the qualifications and assertions of the new technique's developers who understandably advocate its reliability. As was said in People v. King, supra, 266 Cal.App.2d 437, at page 461: ". . . jurors must not be misled by an `aura of certainty which often envelopes a new scientific process, obscuring its currently experimental nature.'" The point was emphasized by our Supreme Court in People v. Collins (1968) 68 Cal.2d 319, at page 332 [ 66 Cal.Rptr. 497, 438 P.2d 33, 36 A.L.R.3d 1176]: ". . . we have strong feelings that such applications, particularly in a criminal case, must be critically examined in view of the substantial unfairness to a defendant which may result from ill conceived techniques with which the trier of fact is not technically equipped to cope."
We turn briefly to an analysis of the technique itself. A sound spectrogram (voiceprint) is produced by a machine known as a spectrograph. It is a basic tool for analyzing speech sounds. When the tape of a voice is properly fed through a spectrograph, a graph is produced which analyzes three main perimeters of speech: time, frequency and intensity of the frequency. The examiner analyzes a spectrogram of the "known" and "unknown" voices to determine identity, i.e., whether the known voice is the same as the unknown voice, or elimination, i.e., whether the known voice is not the same as the unknown voice. Since some characteristics of speech are better recognized by the ear than by the spectrograph, the examiner also listens to tapes when making this determination.
"The sound spectrograph consists of four basic parts: (1) a magnetic recording device, (2) a variable electronic filter, (3) a drum which is coupled to the magnetic recording device and carries a sheet of special paper, [sensitive to] . . . (4) an electric stylus which marks the paper as the drum rotates. The magnetic recording device is first used to record a short sample of speech; the duration of the speech sample corresponds to the time required for one revolution of the drum [ e.g. 2.4 seconds]. The speech sample is then played back over and over again in order to analyze its spectral contents. For each revolution of the drum, the variable electronic filter passes only a certain band of frequencies, and the energy in this frequency band activates the electric stylus so that a straight line of varying darkness is produced across the paper. The darkness of the line at any point on the paper indicates how much energy is present in the speech signal at the specified time within the given frequency band. As the drum revolves, the pass-band of the variable electronic filter moves to increasingly higher frequencies, and the electric stylus moves parallel to the axis of the drum. Thus, a pattern of closely spaced lines is generated on the paper." (Comment, Evidence: Admissibility of Spectrographic Voice Identification, 56 Minn. L.Rev. 1235, 1239; Hecker, Speaker Recognition: An Interpretive Survey of the Literature (1971) Am. Speech Hearing Assn. Monographs No. 16, at pp. 50-51.)
The voiceprint method is based on the fundamental premise that everyone's voice is different. Accordingly, the validity of the technique as a means of personal identification rests on the proposition that the sound patterns produced in speech are unique to the individual and the spectrogram accurately and efficiently displays this uniqueness.
It has been stated that this principal predicate for the soundness of the procedure has never been demonstrated by scientific or empirical data. "The theory of invariant speech is the cornerstone of his hypothesis that individuals can be identified by the spectral characteristics of their voices. The theory posits that the characteristic spectral patterns of phonetically identical utterances vary more between two individuals (intraspeaker variability) than between two such utterances spoken by the same individual (intraspeaker variability). Although so far the theory has not been proven directly, Kersta has buttressed the theory by applying his own hybrid form of statistical probability to the acoustic theory of speech production. He argues that since both the dimensions of the vocal cavities and the coupling of the articulators, which define the spectrum for a given sound, are affected by heredity, sex, age, and socio-environmental factors, it is extremely unlikely that two individuals would develop spectrographically identical speech patterns. While a superficially attractive rationale for voice uniqueness, a proper application of probability theory demands substantially more precision than this." (Fns. 13-17 omitted.) Footnote 18 referred to states: "The number of variables which enter into speech production are not easily defined, either in quantitative or qualitative terms. See HECKER, SPEAKER RECOGNITION at 4-23, 70. In any event, the use of probability theory in court to prove an assumption or conclusion has been criticized by both courts and commentators. [Citations.]" (Comment, The Evidentiary Value of Spectrographic Voice Identification (1972) 63 J.Crim.L.C.P.S. 343, 345.)
Dr. Tosi explained his theory by saying, "Since speech is produced by modulation and resonance of an air stream within the vocal track of a person and this vocal track is as individual as the face or other structural [ sic] of the human body [it] is only logic that these patterns are bias by the personality, [and] the individuality of the person. . . ."
In the case of State v. Cary (1968) 99 N.J. Super. 323 [ 239 A.2d 680], Dr. Tosi testified that the voiceprint method of identification was unreliable and that further experimentation was necessary. Between 1968 and December of 1970 he conducted the study at Michigan State University heretofore referred to. Based upon the results of those experiments, Dr. Tosi changed his position and since that time has, together with Lieutenant Nash, been appearing and testifying in a number of cases in favor of the admission of voiceprint evidence. Since the Tosi experiments are the ones relied upon to justify the admission of the voiceprint method and the testimony in connection therewith, it is necessary to discuss them in some detail.
Dr. Tosi's experiments were conducted over a two-and-one-half-year period at Michigan State University. He used the voices of 250 speakers who represented a sample of the statistical population of 25,000 at the university. All of the speakers selected were about the same age, had the same education, the same midwestern dialect, and had no speech defects or foreign accents. Specifically, there was no experimentation with disguised or mimicked voices. The examiners, who were also university students, were trained to analyze spectrograms for a period of about one month and were paid $2.50 per hour. When the student examiner analyzed the spectrogram, he was required to come to an answer, that is, he was required to either positively identify or positively eliminate. After the examiners analyzed some 50,000 spectrograms, Dr. Tosi compiled the results, which showed that the examiners would misidentify the unknown speaker 6 percent of the time and would erroneously eliminate a known speaker 12 percent of the time. This resulted in an 18 percent error rate, although for purposes of the "forensic model," i.e., falsely identifying the accused, the 6 percent false identification error is the only error factor which is pertinent. In some instances, the examiner was allowed to classify the positive identifications into "almost uncertain," "fairly uncertain," "fairly certain," and "almost certain." In those instances when the examiner indicated he was "certain," a misidentification occurred only 2.4 percent of the time.
According to Dr. Tosi, however, when a professional examiner is used in a real life situation, the error rate in misidentification is "negligible." Dr. Tosi attributed the difference in the error factor between the experimental examiners and the professional examiners to certain differences between the experiments conducted at Michigan State and a professional examiner's evaluation. These differences are: (1) The experimental examiner only examines the spectrogram, while the professional examiner listens to the voices as well, (2) the experimental examiners were always forced to make a decision, while a professional examiner must be absolutely certain before he will testify in court, (3) the examiners used in Dr. Tosi's experiments were given only 15 minutes to make a decision, while a professional takes all the time he needs, (4) the experimental examiners were given either six or nine words of the speaker to work with, while the professional examiner will use all the words he needs or will refuse to come to a decision, (5) the experimental examiner received only one month's training, and (6) the experimental examiners were college students simply participating in an experiment, while the professional examiner bears a great deal of responsibility.
Dr. Tosi concedes that two spectrograms made by the same person uttering the identical phonetic sound on different occasions will not appear the same on the spectrograms because the same person never makes the same sound the same way. However, Dr. Tosi insisted that the difference between two individuals uttering the same sound is always greater than the difference between the same individual uttering the identical sound.
Dr. Tosi admits that there would be some distortion and uncertainty in certain areas of the spectrogram caused by the introduction of a foreign object into the mouth or a throat disease ("distorted articulators"), or by efforts to disguise one's voice. He insists, however, that such distortions would result in false elimination rather than false identification.
Finally, with respect to mimicking and disguised voices, Dr. Tosi was asked if he agreed with Dr. Ladefoged's position that further experimentation was necessary. He replied:
"A. No, essentially he doesn't disagree with me. I think more research is welcome and should be done and I intend to do more research.
". . . . . . . . . . . . . . . . . .
"See, the thing is not completed in all the possible aspects. We have to discard completely. No, there are aspects that are very well-known and then you can use very successfully this method. If we would have a case of mimicking the voice of some other person, two persons would be involved there, and then one person was trying to or there was a suspicion that one was trying to mimic, to immitate [ sic] another person implicated. Then in this case I would think that there would be some kind of reluctance or some doubt.
". . . . . . . . . . . . . . . . . .
"Q. Do you feel that more, as Dr. Ladefoged seemed to indicate, that he felt that one reservation was that he felt more work [was] needed to be done before he would accept, before he, in the disguising of the voice situation?
"A. It would be a case of disguising and you would indicate, yes, in spite of this, yes, maybe, maybe could be some more research done.
In Worley v. State (Fla.App. 1972) 263 So.2d 613, 618, the report of that case indicates Dr. Tosi testified:
"`Q. Did you ever ask any of the 250 persons to attempt to disguise their voices?
"`A. This is another subject. This is another thing we do expect to do and will get more funds from the Department of Justice to do. I have done some experiments on disguised voices. Suppose a defendant makes a telephone call by closing his nose and speaking this way (witness demonstrating), you see. What would happen through a rigorously controlled experiment comparing this voice with the normal? This is what we expect to do. I am certain when criminals know that they can no longer produce a normal telephone voice the first thing they would do is disguise their voices. I would like to know whether they can or cannot. Always they are close, even if a voice is disguised, we are able to find a good percent, but I don't know. I have to produce a very comprehensive and controlled experiment on this matter. This is a good point, a very good point, disguised voices.' (Emphasis added.)"
Doctors Tosi and Ladefoged and Lieutenant Nash readily admitted that the process by which the examiner makes the identification or elimination is highly subjective, that each examiner in a sense follows his own processes and sets his own criteria, and that the reliability and accuracy of the result depends heavily upon the qualification and responsibility of the examiner.
Dr. Ladefoged testified that:
"Q. [By Mr. Raymond] . . . Is there such a requirement as to number of points or similar sounds that the examiner must have before he can say, assuming all his faculties are working correctly, that this is a positive identification?
"A. I don't think that there are any universally agreed set of points of identification or points of similarity that people use.
"This in fact refers to my earlier answer that I couldn't state exactly what it was that Lieutenant Nash was doing when he makes an identification, nor indeed what I'm doing myself.
"I can simply look at the patterns and say they are similiar [ sic]. We have a number of places where we can say, yes, they are similar in this and in this and there are a number of things where we would agree that we would look at certain things.
"But, apart from a few rather obvious ones to those of us in the field, I don't think it's fully clear what are the extra things that we look at."
Some measure of the uncertain and experimental nature of the method, as well as the element of subjectivity, can be found in Lieutenant Nash's testimony. He said that in his experience he had examined 1,500 voices. Of these, he has positively identified 200 to 250, has positively eliminated 600 to 750, and the balance of 500 to 700 were voices he suspected of being the same as the unknown voice but he was uncertain and could not say one way or the other. That is, of the voices examined he had an uncertainty rate of 33 1/3 to 45 percent, an astonishing percentage by comparison to the judicially noticed scientific reliability and certainty of fingerprint evidence.
With regard to the reliability of the technique, Dr. Ladefoged testified that the method is not 100 percent reliable and that Lieutenant Nash's maximum error rate for misidentification is between 5 and 6 percent. "My best estimate of [Lieutenant Nash's] ability is that five or six percent of the time he would say this voice is the same as that voice and be wrong in saying it."
It is interesting to note that Dr. Tosi testified he would not admit the evidence as proof in court if there was a 2.4 percent error. He testified: "Certainly I will do [ sic]. You can imagine that even with 2.4 percent of error of false identification I would not have been in this court today to ask you to admit this kind of evidence even with as [ sic] such reduced error I would not."
Dr. Ladefoged further said sufficient experiments had not been conducted with spectrographs in the areas of speech impairment, foreign objects introduced into the mouth, and women's voices. With respect to disguised and mimicked voices, he said: ". . . we don't know enough about the affects [ sic] of people trying to disguise their voices so that we don't know that somebody, a guilty person, might not go free because he was successful in disguising his voice.
"Nor do we know quite enough about people trying to mimic other people's voices so that we can't be absolutely certain this wasn't somebody trying to frame if that's a legitimate example, trying to frame somebody else for having done something or other."
On the issue of scientific reliability and whether the procedure is sufficiently established to have gained general acceptance in the scientific community, Dr. Ladefoged was equivocal and at best clearly excluded from any suggestion of reliability the cases above mentioned where adequate experimentation had not been done.
Dr. Ladefoged testified:
"Q. [By Mr. Levis] And do you now feel that the voiceprint is a scientifically accepted method of identification of voices?
"A. I'm not quite sure what you would mean by scientifically acceptable. It is a method which can give a great deal of indication as to whether two voices are the same or not the same.
"I don't understand the term `scientifically acceptable' in that sense.
". . . . . . . . . . . . . . . . . . . . . .
"Q. And basically then could you describe what it is that he [Lieutenant Nash] does, and I will give you a hypothetical example.
"If two tape recordings were submitted to Lieutenant Nash and one was the voice of an identified person and the other was the voice of an unidentified person, through the use of the spectrograph could he make a comparison of the voices and reach a conclusion as to whether or not they were in fact the same person?
"A. Given the fact that the recordings were satisfactory and given the fact also that he does more than just look at the pictures, he also listens to the two tape recordings, he listens to one of the known voice and of the unknown voice, both by looking and listening together, yes, he could make a good estimate as to whether they were the same voice or not.
". . . . . . . . . . . . . . . . . . . . . .
"Q. Is it then, sir, your testimony that at this time you feel that the use of the voice spectrograph in the identification of or comparison of known voices with unknown voices is a scientifically acceptable method for identification?
"A. Given certain safeguards so that it's performed by a process or that the recordings that the investigator has truely [ sic] identified and of no noises or background, undue degree of background noise, given also the fact that there is no claim that these are women's voices of high pitch which make it difficult, also that there is no claim that this is somebody mimicing [ sic] somebody elses voice, the defendants, that is somebody mimicing and you can say there is a skilled mimic who is able to mimic very well, then I think that's most of the circumstances, yes, and, oh, and also given the fact that the court is advised of the fact that this is not 100 percent, yes, I would agree that this is a technique that is useable [ sic] in a court of law."
Dr. Tosi, on the other hand, said the procedure was scientifically valid and accepted by the scientific community in that his experiments have "answered most of the questions posed by the scientific community." (Italics added.)
The results of Dr. Tosi's Michigan study were published in The Journal of the Acoustical Society of America in 1972 (Tosi, Oyar, Lashbrook, Pedrey, Nichol Nash, Experiment on Voice Identification, 51 J. Acoust. Soc. Am. 2030-2043). The trial in the cause at bench was held between May 8, 1972 and May 18, 1972; obviously, insufficient time transpired between the publication and the trial for widespread analysis and reaction of the scientific community to the Tosi experiments. The same observation can be made about the other reported cases in which Dr. Tosi and Lieutenant Nash testified and in which the testimony was admitted. ( Hodo v. Superior Court, supra, 30 Cal.App.3d 778; State ex rel. Trimble v. Hedman (1971) 291 Minn. 442 [ 192 N.W.2d 432]; Worley v. State (Fla.App. 1972) 263 So.2d 613; Alea v. State (Fla.App. 1972) 265 So.2d 96; see also State v. Andretta (1972) 61 N.J. 544 [ 296 A.2d 644].)
Apparently, reports of Dr. Tosi's study were available in 1971 through the Department of Audiology and Speech Sciences at Michigan State University. (See United States v. Addison (1974) 498 F.2d 741, 744, fn. 7.) These reports, however, were apparently not widely circulated, and the article referred to above was the one which prompted the widespread criticism hereinafter referred to.
In fact, the case of United States v. Raymond (D.D.C. 1972) 337 F. Supp. 641, which was relied on by the prosecution in the instant case and in Hodo v. Superior Court, supra, was recently disapproved in United States v. Addison, supra, 498 F.2d 741, where the court placed considerable reliance upon the recent plethora of articles and papers from highly respected authorities condemning the procedure as presently unreliable, as being unaccepted in the scientific community, and as being insufficiently tested to justify the imprimatur of the scientific community.
The following excerpts from a few of these publications will suffice to illustrate the scientific community's refusal to accept Dr. Tosi's claims:
Bolt, Cooper, David, Denes, Pickett Stevens, Speaker Identification by Speech Spectrograms: Some Further Observations (1973) 54 J. Acoust. Soc. Am. 531: "Our interpretations of the new data [Tosi study] lead us to reiterate our previous conclusion: that the degree of reliability of identification under practical conditions has not been scientifically established."
The former conclusions referred to are contained in an article entitled Speaker Identification by Speech Spectrograms: A Scientists' View of its Reliability for Legal Purposes (1970) 47 J. Acoust. Soc. Am. 597, by the same authors, which, after summarizing the scientific objections to the use of spectrograms for speaker identification under seven separate categories, concluded: "We find, in brief, that spectrographic voice identification has inherent difficulties and uncertainties. Anecdotal evidence given in support of the method is not scientifically convincing. The controlled experiments that have been reported give conflicting results. Furthermore, the experiments reported thus far do not provide a direct test of the practical task of determining whether two spoken passages were uttered by the same speaker or by two different speakers, one of whom may be a person unknown.
"We conclude that the available results are inadequate to establish the reliability of voice identification by spectrograms. We believe this conclusion is shared by most scientists who are knowledgeable about speech; hence, many of them are deeply concerned about the use of spectrographic evidence in the courts. Procedures exist, as we have suggested, by which the reliability of voice identification methods can be evaluated. We believe that such validation is urgently required."
Hazen, Effects of Differing Phonetic Contexts on Spectrographic Speaker Identification (1973) 54 J. Acoust. Soc. Am. 650: "The data presented here indicate that the forensic value of sound spectrograms for speaker identification purposes is quite limited. It was suggested earlier that comparison of speech samples of sufficient duration (a yet undetermined parameter) in identical contexts might serve to provide an indication of a speaker's identity. Thus spectrographic speaker identification may have some use as an investigative tool. As stated above, this is quite different from using spectrograms for absolute identification. Furthermore, most applications would occur under essentially open test conditions, a factor which has been shown to decrease reliability.
"The data also indicate that spontaneous speech itself further decreases reliability. To the author's knowledge, the present study is the only study of spectrographic speaker identification to utilize spontaneous speech. It is reasonable to demand that data which is relied upon to establish the reliability of this technique in forensic applications be collected utilizing spontaneous elicited conversation.
"Particular note should be made of relevant variables not investigated in the present study or in other studies, e.g., attempts at speech disguise, as well as others mentioned above.
"With these factors in mind, it is recommended that sound spectrograms not be used for speaker identification purposes until sufficient and consistent data are gathered to establish accurately the limits of this technique." (At p. 659.)
Comment, The Evidentiary Value of Spectrographic Voice Identification (1972) 63 J.Crim.L.C.P.S. 343, in which the writer concludes at page 355: ". . . the courts ought not to be too eager to embrace a technique which has only recently been rescued from an abyss of scientific scorn — a rescue accomplished by the narrow experimental data produced in one study [Tosi study]."
See also: Poza, Voiceprint Identification: Its Forensic Application, Stanford Research Institute, published in The Proceedings of the 1974 Carnahan Crime Countermeasures Conference, University of Kentucky; Note, Voiceprint Identification (1972-1973) 61 Geo.L.J. 703, at pages 715, 724-727, 744-745; Jones, Danger — Voiceprints Ahead (1973) 11 Am.Crim.L.Rev. 549, particularly footnote 120 at page 569, where the author lists a large number of scientists who continue to have misgivings about the reliability of the technique. (3a) Thus, Dr. Ladefoged's reluctance to express unconditioned acceptance of the voiceprint method and his effort to hedge and qualify his statements regarding the acceptance of it in the scientific community are understandable in light of these references. His testimony together with the above references demonstrate that the scientific reliability of these procedures is far from established or accepted by the members of the areas of scientific expertise most directly connected with the relevant disciplines.
Even if we accept at face value Dr. Tosi's position that his Michigan State project was well designed and carried out in a scientifically objective manner and that the study developed sufficient new data to establish the reliability of the voiceprint identification technique, it is admitted by both Dr. Tosi and Dr. Ladefoged that the experiment did not include trials with disguised or mimicked voices and that additional experimentation in these areas is desirable. (4) It is well established that a fundamental prerequisite to the introduction of any evidence regarding a scientific experiment is that the experiment relied upon must have been conducted under conditions substantially similar to those present in the occurrence which is the subject of the litigation. ( People v. Roberts (1953) 40 Cal.2d 483, 491 [ 254 P.2d 501]; People v. Orchard (1971) 17 Cal.App.3d 568, 577 [ 95 Cal.Rptr. 66]; People v. Stuller (1970) 10 Cal.App.3d 582, 599 [ 89 Cal.Rptr. 158, 41 A.L.R.3d 712], cert. den. 401 U.S. 977 [28 L.Ed.2d 327, 91 S.Ct. 1205]; see also People v. Busch (1961) 56 Cal.2d 868, 878 [ 16 Cal.Rptr. 898, 366 P.2d 314].) (3b) It necessarily follows that Dr. Tosi's testimony pertaining to the general acceptance of the procedure in the scientific community and relating to the small percentage of error in identification of unknown voices based upon those experiments carries little weight in this context. For it must be remembered that in giving samples of his voice to investigating officers, appellant was directed to and did mimic the recorded voice of the bomb threat caller. Considering this factor, we are not satisfied that the results of Dr. Tosi's experiments warranted the admission of the voiceprint evidence.
Accordingly, the conclusion is compelled that with respect to disguised and mimicked voices in particular, the prosecution did not carry its burden of proof to demonstrate that the scientific principles pertaining to spectrographic identification were beyond the experimental and into the demonstrable stage or that the procedure was sufficiently established to have gained general acceptance in the particular scientific field in which it belongs. ( Huntingdon v. Crowley, supra, 64 Cal.2d 647, 653-654.) On the facts of this case, the voiceprints and testimony of Lieutenant Nash should have been excluded from evidence, and appellant's conviction of the count relating to the bomb threat made to the police predicated upon that evidence must be reversed.
In arriving at this conclusion, we fully understand and appreciate the difficulty of proving the identity of a caller making an anonymous threatening or annoying call. This especially troublesome problem of proof, coupled with the diabolical nature of the crime, would make it easy for the courts to relax the stringent standards of proof required in all criminal cases and admit evidence that would not otherwise be permitted. (5) It is our duty, however, where the life or liberty of a defendant is at stake, to be particularly careful that there is not only substantial evidence to support the implied finding of identity but that the finding is based upon admissible and nonprejudicial evidence. (6) "A defendant's identity must be established beyond a reasonable doubt. When identification is chiefly founded upon an opinion which is derived from utilization of an unproven process or technique, the court must be particularly careful to scrutinize the general acceptance of the technique." ( People v. King, supra, 266 Cal.App.2d 437, 459.) Our standards of justice for court proof require nothing less.
The usefulness of this relatively new technique, of course, is not lost and its employment as a valuable investigative aid will continue. Indeed, it appears that the California Department of Justice takes this position. In the May 1973 Peace Officer Law Report it is stated that "At the present time, results of voice comparisons will be offered as investigative aids only and court testimony will not be provided."
As the now famous expatriated Russian author Aleksandr I. Solzhenitsyn said in The First Circle (1968) pp. 197-198): "This was a new science — finding a criminal by a print of his voice.
"Until now they had been identified by fingerprints. They called it dactyloscopy, study of the finger whorls. It had been worked out over the centuries.
"The new science could be called voice study — that was what Sologdin would have called it — or phonoscopy. And it all had to be created in a few days."
THE TWO MISDEMEANOR CONVICTIONS OF VIOLATING PENAL CODE SECTION 653m, SUBDIVISION (a)One of the violations of Penal Code section 653m, subdivision (a), is based on a call received at about 12:30 p.m. on June 11, 1971, by Jessie Faretta, a telephone operator at St. Agnes Hospital. The male caller spoke with a Spanish accent and inquired, "You have a Georginia Law working there." Mrs. Faretta replied, "I will pass you to the personnel office." The caller responded, "Oh no — I — do you have someone who speaks Spanish?" Mrs. Faretta replied that she could understand him, whereupon the caller said, "Ed Law killed my friend Manuel Chico two days ago and you have a Georginia Law working there and we're going to bomb Saint Agnes." Mrs. Faretta thought that the Spanish accent might have been assumed.
The other misdemeanor conviction is based upon a call received by Barbara Bertolli, a switchboard operator at St. Agnes Hospital, at about 7:06 p.m. on July 17, 1971, from a person who spoke with "an accent" and said, "You and Georginia Law are going to be blowed all to hell tonight."
Appellant was charged with two additional counts upon which he was acquitted: One was as a result of a telephone call made January 7, 1972, at approximately 6:15 p.m. to Susan Patterson, an employee of Bob's Big Boy Restaurant, from a man who sounded like a Negro and who said, "There's a bomb planted in your restaurant and it's going to blow up within twenty-four hours." The other count was founded upon a telephone call made January 7, 1972, at approximately 5 p.m. to Joyce Matula, an employee of the Bank of America, Blackstone-Bremmer Branch, in which the male used a "put-on Negro voice" and said, "You have three days to get out of the bank before it's blown to bits. If you think what's going on in New York is bad you ain't seen nothing yet. Do you understand me? Do you hear me?"
The trial court permitted the prosecution to introduce evidence of many other threatening calls in the Fresno area, none of which formed the basis of charged offenses. In sum, these were:
A call received by Leslie J. Mayer, a switchboard operator at St. Agnes Hospital on July 5, 1971, at approximately 10:45 a.m., in which a man who talked with a Southern drawl said, "Georginia Law is still working there and I'm going to blow you up in two weeks."
Another was a call received by Joan Nikitan, a switchboard operator at St. Agnes, from a man who spoke with a southern or Negro accent and said, "If Georginia Law is working at the hospital any time during the next week we will blow you to hell like New York." Mrs. Nikitan thought that the accent might have been assumed but was unsure. The call was received on January 7, 1972, at approximately 4:40 p.m.
On January 12, 1972, at about 5:21 a.m., Carol Hatfield, a switchboard operator at St. Agnes Hospital, received a telephone call from a man with "a southern or Tennessee accent," who stated, "If you don't want to see your hospital go up in ashes and smoke, you had better get rid of Georginia Law within the next three days or it will be burned with a bomb."
On May 23, 1971, Luther Hendrix, Jr., a security guard at the Fresno Bee, received two telephone calls from a person he described as 45 to 50, Caucasian, with a clear voice. The first call was received at 6:05 a.m. and the caller stated, "We're going to kill Ed Long in approximately thirty minutes and there's nothing you or anyone else, anyone can do to prevent it. And then we're going to hit the Federal Building."
The same man called a second time at 6:35 a.m. and stated, "Ed Long is now dead and in forty-five minutes the Fresno Bee gets it. And there's nothing you or anyone else can do about it."
On July 17, 1971, at 10:24 p.m. Louise Bridges, a switchboard operator at the Fresno Police Department, received a call from a male with a "southwestern, country western type accent," who stated, "If the police don't stop protecting Ed Law we're going to blow up the police headquarters."
On July 17, 1971, Arnold Fiorani, proprietor of the Bel Air Motel in Fresno, received two telephone calls wherein the caller mentioned appellant's name in connection with a threat. On that date appellant was a registered guest at the motel. The first call was at approximately 1:30 p.m., from a male who spoke with some sort of an accent and asked if Ed Law was registered there. Mr. Fiorani replied that he was, whereupon the caller stated, "If you don't get him out the motel will be bombed." About 45 minutes later the caller phoned again and said, "This is your last warning, if you don't get Ed Law out the motel will be bombed."
With the exception of the call to Carol Hatfield at St. Agnes Hospital on January 12, 1972, there was no direct proof that any of these charged or uncharged calls was made by appellant, in that the recipients were either not asked to or were unable to identify the voice of the appellant as being that of the unknown caller. With respect to the call to Carol Hatfield, she was able to identify the voice of appellant as that of the unknown caller from one of four voices played to her in court.
Appellant made motions to strike the evidence of all of the uncharged calls on the ground they were irrelevant because there was insufficient evidence connecting appellant to these calls. The motions were denied.
The prosecution's theory was that the uncharged calls were admissible to show motive and a modus operandi by appellant to prove identity of the appellant as the one who made the Faretta and Bertolli calls to St. Agnes Hospital and the calls for which appellant was acquitted. (Evid. Code, § 1101, subd. (b); People v. Archerd (1970) 3 Cal.3d 615, 638-639 [ 91 Cal.Rptr. 397, 477 P.2d 421].) (7) Evidence of other crimes when sufficiently connected with a defendant is evidence of a defendant's identity as the one who committed the crime charged. ( People v. Poulin (1972) 27 Cal.App.3d 54, 65-66 [ 103 Cal.Rptr. 623].)
(8a) The one call made to St. Agnes Hospital which was received by Carol Hatfield and which was independently established as criminal conduct by appellant through her identification of his voice, coupled with the identical motive and similar modus operandi, would make evidence of this call admissible under the theory urged; similarly, the two other uncharged calls to St. Agnes Hospital in which Georginia Law was mentioned had sufficient common identifying features with the charged calls to render them admissible in the discretion of the trial court. Taken together, this evidence is sufficiently substantial to support the finding of identity of appellant as the one who made the calls to Mrs. Faretta and Mrs. Bertolli at St. Agnes Hospital. ( People v. Redmond (1969) 71 Cal.2d 745, 755-756 [ 79 Cal.Rptr. 529, 457 P.2d 321].)
However, as to the other five calls (two to Mr. Hendrix at the Fresno Bee, two to Mr. Fiorani at the Bel Air Motel, and one to Mrs. Bridges at the police station), there was not an identity of motive, and similarities in the modus operandi, if any, were weak. (9) The reason for normally excluding such evidence aside from the patently prejudicial nature thereof, is ". . . to avoid placing the accused in a position where he must defend against crimes for which he has not been charged and to guard against the probability that evidence of other criminal acts having little bearing on the question whether defendant committed the charged crime would assume undue proportions and unnecessarily prejudice defendant in the minds of the jury, as well as to promote judicial efficiency by restricting proof of extraneous crimes." ( People v. Cramer (1967) 67 Cal.2d 126, 129 [ 60 Cal.Rptr. 230, 429 P.2d 582].)
"Several decisions have held that the test of admissibility of evidence of another offense offered to prove common design, plan, or modus operandi is whether there is some clear connection between that offense and the one charged so that it may be logically inferred that if defendant is guilty of one he must be guilty of the other. Or as the matter is sometimes stated, the other offenses offered to prove pattern, scheme, or plan are sufficiently similar and possess a sufficiently high degree of common features with the act charged where they warrant the inference that if the defendant committed the other acts he committed the act charged." ( People v. Cramer, supra, 67 Cal.2d 126, 129-130; see People v. Poulin, supra, 27 Cal.App.3d 54, 65-66.)
(8b) Under this test, we have examined these five uncharged calls and find them wanting, there being a lack of common identifying characteristics between those calls and the two misdemeanor counts upon which appellant was convicted clearly connecting appellant with both of the offenses. (See People v. Albertson (1944) 23 Cal.2d 550, 578, 580-581 [ 145 P.2d 7].) The offenses offered to prove the pattern, scheme or plan are not sufficiently similar nor do they possess a sufficiently high degree of common features with the acts charged to warrant the inference that if appellant committed the uncharged acts he must have committed the charged acts. We therefore hold that it was error to admit evidence of these five uncharged offenses.
We conclude, however, that the error was not prejudicial. It is to be noted that with the exception of the bomb threat call to Officer Holland, the jury convicted appellant only of the two counts relating to calls to St. Agnes Hospital and acquitted the appellant of making calls to Bob's Big Boy Restaurant (count two) and the Bank of America (count five). Since the five uncharged calls did have much in common with the two counts upon which appellant was acquitted, the fact of the acquittal clearly indicates that the jury did not consider the five uncharged calls of significant probative value with relation to those two counts and, a fortiori, of no weight with relation to the conviction of making the two calls to St. Agnes Hospital. It would appear, therefore, that the jury concluded that the appellant was guilty of the calls to St. Agnes Hospital upon the basis of the evidence of the other calls that were made to the hospital, and ignored the five uncharged calls. We have concluded therefore that it is reasonably probable that a result more favorable to the appealing party would not have been reached if the error in admitting evidence of the five uncharged calls had not been committed, and therefore the admission of that evidence was nonprejudicial. ( People v. Watson (1956) 46 Cal.2d 818, 836-837 [ 299 P.2d 243].)
The judgment is reversed as to count one and affirmed as to counts three and four.
Gargano, J., and Franson, J., concurred.