Opinion
No. 71-527.
May 19, 1972. Rehearing Denied July 17, 1972.
Appeal from the Criminal Court of Record, Orange County, Warren H. Edwards, J.
Timothy A. Curran of Law Office of Sam E. Murrell Sons, Orlando, for appellant.
Robert L. Shevin, Atty. Gen., Tallahassee, Nelson E. Bailey, Asst. Atty. Gen., West Palm Beach, and Rom W. Powell, County Sol. of Orange County, Orlando, for appellee.
Defendant was convicted of telephoning false bomb threats. The evidence against him included voiceprints. Their admission is assigned as error. We find voiceprints were competent evidence and were properly admitted. We affirm.
The police received two anonymous phone calls that bombs had been placed in the police station and in a utilities building. The second call was traced to a phone booth. Defendant was apprehended nearby.
Defendant's fingerprint was on the phone. He gave conflicting stories concerning his presence in the phone booth.
Defendant was also identified by voice by the officer receiving both calls. Such voice identification by ear is properly admissible in Florida. Cason v. State, Fla.App. 1968, 211 So.2d 604; Simon v. State, Fla.App. 1968, 209 So.2d 682 and Weinshenker v. State, Fla.App. 1969, 223 So.2d 561. The credibility of such evidence is clearly a jury question.
In addition, a tape was made of defendant's voice and it was sent, along with tapes of the bomb threats, for spectrographic analysis, to the voiceprint unit of the Michigan State Police Department. There, spectrograms charting the patterns of speech sounds were made of both tapes. These voiceprints were matched to determine if they were made by the same voice.
At trial Dr. Tosi, Professor of Audiology and Speech Sciences, Michigan State University and his collaborator, Sergeant Nash, of the Michigan State Police Department, both testified at length concerning the reliability of voiceprints. They testified, based on their own extensive experiments with over 34,000 spectrographic comparisons, and on the works of others. They claimed in excess of 98% of the time trained examiners could reliably match voices by comparing voiceprints.
After being accepted as qualified experts by the court the witnesses demonstrated how they were sure the voiceprint of the caller and defendant were the same.
Florida courts have long enjoyed considerable discretion in the admittance of novel or experimental evidence, if they feel certain standards of scientific reliability have been attained. See Coppolino v. State, Fla.App. 1968, 223 So.2d 68. We feel such standards prevailed here.
The admission of voiceprint evidence is a case of first impression in Florida. In fact, our research indicates only five previous appellate decisions nationally on this point, two refusing to admit voiceprints and three admitting them.
State v. Cary, 1967, 49 N.J. 343, 230 A.2d 384, was the first. There the New Jersey Supreme Court concluded that "as of this date" voiceprints have not attained "such degree of scientific acceptance and reliability as to be acceptable as evidence." Dr. Tosi testified that more experimentation was needed, but he had begun such experiments, which would be completed in two years. See also State v. Cary, 1968, 53 N.J. 256, 250 A.2d 15; State v. Cary, 1969, 99 N.J. Super. 323, 239 A.2d 680 and State v. Cary, 1970, 56 N.J. 16, 264 A.2d 209.
In People v. King, 1968, 266 Cal.App.2d 437, 72 Cal.Rptr. 478, it was also decided that a "sufficient level of scientific certainty" has not been reached to allow voiceprints to be admitted into evidence for identification purposes.
Both of these cases are over three years old. Since then, impressive scientific data has been amassed as to the voiceprint's reliability. We hypothesize that, based on the changes in Dr. Tosi's testimony alone, Cary and King would be decided differently today. For these reasons, we decline to follow these cases.
The first case holding voiceprints admissible was United States v. Wright, 1967, 17 U.S.C.M.A. 183, 37 C.M.R. 447. Voiceprints were held admissible only for corroboration purposes.
The next case is State ex rel. Trimble v. Hedman, Minn. 1971, 192 N.W.2d 432. This excellent opinion by Justice Knutson discusses at scholarly length the development and use of voiceprints. Hedman also involved the testimony of Professor Tosi. By this time his experiments were complete and he unequivocally, as he did here, testified as to the scientific accuracy and reliability of voiceprints. Hedman concluded that voiceprints are admissible to corroborate voice identification by ear, if a proper foundation is laid establishing the expertise of one preparing the spectrograms.
The last case is United States v. Raymond, United States District Court for the District of Columbia, February 2, 1972, 337 F. Supp. 641. The court went into great detail discussing the scientific basis for voiceprint comparisons. It also commented upon the various experiments conducted, their failings and successes, in concluding voiceprint comparisons were admissible if the expertise of the examiner was established. They decided their expert, who was also one of ours, Nash, had the requisite expertise.
In our case the evidence against defendant was already ample to sustain his conviction, even without the use of voiceprints. Therefore, this decision must be limited by our facts. We hold voiceprints were properly admitted to corroborate defendant's identification by other means. State ex rel. Trimble v. Hedman, supra.
The issues not being before us, we do not decide if voiceprint identification may be employed only for corroboration, or, if voiceprint identification, standing alone, would be sufficient to sustain the identification and conviction of the defendant.
Affirmed.
MAGER, J., concurs specially.
WHITE, JOSEPH S., Associate Judge, dissents.
I fully concur with the opinion of Judge Walden for those reasons which he has so ably expressed. In addition to those views, it is my further view that the judgment of conviction can be sustained even on the basis of the voiceprint testimony alone.
I fail to discern a distinctive difference between the analysis and identification of a voice by an expert based solely upon the scientific reproduction thereof (i.e. voiceprints) and the identification of a voice by a lay witness based merely upon hearing the voice. Simon v. State, Fla.App. 1968, 209 So.2d 682. It would seem that in each instance the question becomes one more properly relating to the weight or value to be given to such identification or testimony by the trier of fact. Clearly the victim of a threat or an obscene call is permitted to testify as to the identification of such voice by comparison with that of the alleged perpetrator. See Cason v. State, supra; Weinshenker v. State, supra; and Simon v. State, supra. See also annotation in 24 A.L.R.3d 1261.
As the Supreme Court of California in People v. Ellis, 1966, 65 Cal.2d 529, 55 Cal.Rptr. 385, 421 P.2d 393, observed in holding that voice identification testimony is not within the privilege against self-incrimination:
"Voice identification testimony is the product of an observable physical characteristic made by an independent witness. It is the very type of objective factual evidence, independent of information communicated by the accused, that the privilege encourages police to seek. . . ." (Emphasis added.)
What I am suggesting is a recognition that voice identification, based upon scientific techniques approaching the character of experimentation reflected in this case, ought to be admissible into evidence as "direct evidence" in addition to its admissibility as corroborating evidence. Cf. State ex rel. Trimble v. Hedman, supra. In my view the character of this evidence (once the proper predicate has been laid as to the expertise of those involved and the method of analysis and experimentation as was done in this case) hurdles the question of admissibility falling more properly within the ambit of probative value which is a question of fact for the jury to decide. Cf. Mack v. State, 1907, 54 Fla. 55, 44 So. 706.
It should be noted that the same persons who testified here were the same individuals who testified in the trial below in connection with voice identification also testified in State ex rel. Trimble v. Hedman, supra, the court there commenting that "the qualifications of all of these experts are hardly open to question".
Doubtless there are those who will question the reliability of voice spectrograms and its general scientific acceptance. See State v. Cary, supra. Cf. State ex rel. Trimble v. Hedman, supra. However, the accuracy of these experiments has reached the degree where its admissibility should be less open to question. In this context, a comprehensive discussion of spectrogram voice identification appears in 19 Am.Jur., Proof of Facts Annotated, p. 431, where the text writer observes:
"A number of subjective experiments have been conducted with voice spectrograms that show a high degree of identification success. In these experiments, numerous voiceprints were made of the same words spoken by different persons. Each utterance of a word was recorded on a separate card. The cards were shuffled and trained persons were instructed to group them according to the voice they represented. Out of the nearly 50,000 decisions, these trained people made the correct decision more than 99 percent of the time."
Again, I should stress that admissibility should not be confused with credibility. The fact that voiceprint testimony may be admissible does not mean that the jury is categorically bound by such testimony or is in any way restrained from being able to determine the proper weight to be given to such testimony. My faith in the jury system leads me to believe that it will be given the weight that the situation and circumstances may dictate. Furthermore, voiceprint testimony should be no less admissible and no less credible than the testimony of a lay witness or a victim who is permitted to make an identification from merely having heard the voice of the accused. Simon v. State, supra.
In this day and age where crime rates approach alarming proportions and where recent decisions of the United States Supreme Court establish stringent guidelines in the investigative, custodial and prosecutional areas a premium is placed upon the development and use of scientific methods of crime detection. Protecting society from those who have violated the law as well as protecting the one who has been unjustly accused serves to heighten the need for more sophisticated methods of crime prevention and crime detection. The use of spectrograph voice identification or "voiceprints" can become an effective tool in identifying the perpetrators of telephone bomb threats, extortioners, kidnappers, as well as persons making obscene telephone calls.
Correspondingly, scientific methods of voice identification can serve to virtually eliminate those instances where the wrong person has found himself unjustly charged or under suspicion.
In my view, voiceprint identification "has passed the boundary from the experimental to the status of a demonstrably valid investigative and forensic tool", 19 Am.Jur., Proof of Facts, p. 438, to be utilized as "direct" as well as "corroborating" evidence. See also United States v. Raymond, supra.
I find myself unable to agree that use of the "voiceprint" testimony was proper.
The term "voiceprint" is of recent origin, and, so far as I have been able to discover, has not found its way into any dictionary. Its use has been criticized as "implying a misleading analogy to fingerprints." See Hello, This is * * * Use of "Voiceprints" In Court Proceedings Creates Legal Debate, The Wall Street Journal, March 13, 1972, p. 1, col. 1.
It appears from the record before us that a "voiceprint", in fact, is a spectrogram, made by imprinting a person's voice upon a magnetic spectrograph tape. This is placed upon a "marking drum" in a spectrograph, which records a permanent pattern of the person's voice on the spectrogram, so that it can be visually examined. In determining whether or not two voices, previously recorded on tapes, are the same, the examiner must not only visually examine the spectrogram, but also must listen carefully to the sound of the recorded voices. Thus, the senses of sight and hearing are involved in forming the necessary opinion.
For a much more detailed explanation of the technique involved in the "voiceprint" process see State v. Cary, (1968), 99 N.J. Super. 323, 239 A.2d 680; People v. King, (1969) 266 Cal.App.2d 437, 72 Cal.Rptr. 478; and, United States v. Raymond, D.C. (1972), 337 F. Supp. 641.
In the case at bar, one of the tapes tested on the spectrograph was recorded on a telephone recorder. The other was of defendant's voice recorded on an ordinary tape recorder.
The recording of voices on magnetic tapes, likewise, is of comparative recent origin. In this process sound is converted to electrical energy which in turn sets up a magnetic pattern on the tape. During playback this pattern on the tape is reconverted to electrical energy, amplified by electronic process, and changed back to sound. It was stated at the trial that one who testifies as an expert in this field must be trained, not only in phonetics and spectrology, but in electronics as well.
The State of Florida produced at the trial a witness, Dr. Oscar Tosi, a well-known proponent of the "voiceprint" technique, and a frequent witness in trials involving the subject. He was brought from Michigan State University, where he is a Professor in the Department of Audiology and Speech Sciences and Pathology. He stated that he held a doctorate in Physics and Engineering from Buenos Aires University and another in Audiology and Speech Sciences and Electronics from Ohio State University. He stated that he had made many experiments in the field in question under a grant of $300,000.00 from the United States Government.
So far as is disclosed by the record he had not, himself, examined the tapes involved in this case. His testimony concerned solely the work he had done in the field, and his opinions regarding the integrity of the process and the qualifications of another State's witness, Sergeant Nash likewise brought from the State of Michigan to testify at the trial.
It was Sergeant Nash who had made experiments in the State of Michigan with the tapes involved in the trial below. From these, he testified that, in his opinion, the voice on each tape was that of the defendant. So far as the record discloses Sergeant Nash had never heard the defendant's voice, except as it may have been recorded on the tape sent to him by local police.
The defendant took the stand and stated, categorically, that he had not made the telephone call to the police station.
It is from the testimony of Dr. Tosi that I conclude that the use of the "voiceprint" process at the trial should not be approved, because of lack of reliability.
First, there is the margin of error in making false identification.
In support of the reliability of the "voiceprint" process, reference has been made to the following statement from 19 Am.Jur., Proof of Facts, p. 431: "Out of nearly 50,000 decisions, these trained people made the correct decision more than 99 percent of the time." The author of this statement neglects to give the source from which this information was obtained so that it is impossible to check the accuracy of the statement. At any rate, the record before this court contradicts the statement.
Dr. Tosi readily conceded that the procedure is not 100 per cent accurate. In making 34,996 tests he "found some five per cent of false identification." By eliminating those who were "uncertain" and could make no identification he "reduced the five percent to two percent." He testified:
"And keep in mind that a professional examiner is not forced to give a decision, not in all occasions. It has happened to me on many occasions that I have said, `Well, I am unable to give the conclusion. The samples are not good.' Or any reason. So, I don't know. So this forced — they gave us their judgment and I applied this scale to my cases. So we should conclude that actually they were two percent that they were wrong and they were convinced that they were right, so this is a pure, let's say two percent of error out of this general five percent of error that I reported to you here in Court because I have to report the truth."
There are approximately 333 million English speaking people in the world. See The World Almanac, 1972, p. 334. When we think of the number of fingerprints made in verifying the accuracy of fingerprint identification, 34,000 tests seem a rather negligible number to accept as proof that no two voices among 333 million people are alike.
An important factor in fingerprint identification lies in the fact that they can be classified. Dr. Tosi, himself, recognizes the importance of such a factor. He testified:
"We do not classify voices. In my dream, it is my dream to have a file of voices and when you have a voice of the offender you go to the file and select ten or fifteen that have the same classification and then try to see if they are known within this or not. The same as fingerprints. In the context, yes, I think more studies have to be done." (Emphasis added.)
Also, it appears from the testimony of Dr. Tosi that another important feature of this process is yet to be studied, namely, the problem of the disguised voice, or mimic. The ability of one person to mimic another is seen and heard frequently on television programs. Here is what Dr. Tosi says in this regard:
"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.)
Thus it appears that it has not yet been determined that a person cannot defeat the procedure by disguising his voice. For this reason, the witness today cannot say with certainty that defendant's voice was not, by way of disguise, imitated by another person in making the telephone call involved here.
This demonstrates quite clearly that this technique is still in its infancy and that experiments are still in progress which might change the complexity of the whole matter. For instance, let us imagine that in the completion of studies now under way and in the execution of experiments to be made when further funds are available, it is discovered with a mimic can so disguise his voice that the disguised voice cannot be distinguished from the real voice of the person imitated. Such seems quite possible according to the testimony found here, and Dr. Tosi feels certain that criminals will take advantage of such a deficiency. Obviously, the value of "voiceprint" testimony in a criminal trial then will be of little value.
This is a critical defect which exposes the present state of this process as a dangerous and unsafe mode of proof in a criminal trial. Neither the case of State ex rel. Trimble v. Hedman, Minn. 1971, 192 N.W.2d 432, nor United States v. Raymond, supra, — decisions in which "voiceprint" testimony was held to be admissible evidence, — comments on this phase of the subject. It is possible, of course, that the testimony before the courts making those decisions did not disclose the uncertainty of the process in this regard, as does the testimony of Dr. Tosi in the case at bar and which is quoted above.
Few will dispute the statement that in this day and time, crime in our country is rampant. In many of our cities public streets and places are unsafe for travel after dark. Private property rights are flagrantly violated and trespassed upon. Respected police officers who twenty-four hours a day place their lives on the line for the citizens of this country, are frequently subjected to insults and indignities, to say nothing of assassinations and physical injury. There is a widespread disregard for law and order among those out to overthrow the "establishment". Law abiding citizens are alarmed, and rightly so. However, guilt of one accused of crime still must be proven beyond a reasonable doubt. The courts still are charged with the duty of protecting the innocent, as well as punishing the guilty. When it comes to human liberty the only grade allowed should be 100 per cent accuracy. This had been the policy of the courts in Florida in the past. See 13 Fla.Jur., Evidence, Sec. 125. We must be cautious not to relax it now.