Phil Rose

Ph.D. (Cambridge 1982)
M.A. (Manchester 1974)

B.A. Hons First Class (Manchester)
Dip. IPA First Class (London)

I am a speech scientist with expertise in Forensic Voice Comparison and the phonetics of Asian Tone Languages, like Chinese and Thai. I am also a professional forensic speech science consultant.

I am a member of the Australian Academy of Forensic Sciences, and past chairman of the Forensic Speech Science Committee of the Australasian Speech Science and Technology Association. I have also been a British Academy Visiting Professor at the Joseph Bell Centre for Forensic Statistics and Legal Reasoning at the University of Edinburgh.

These pages contain links to my forensic and tonal papers; to Chinese tonal data that you can listen to; to some of the numerical acoustic data, much of it multispeaker, that I have used in my research; and to some first-hand recordings of a member of the elite WW2 German Brandenburger commando unit forming the basis of the novel-memoir Iron Crossed. And there is also a bit of a bio.

Last update: 19-apr-24


In forensic voice comparison the expert typically compares suspect and offender speech samples - ultimately to help interested parties decide whether the suspect said the incriminating speech.

In the late 1990's I pioneered the application of the Likelihood Ratio of Bayes' Theorem to traditional forensic voice comparison, thus bringing it into line with forensic DNA profiling. Likelihood Ratios are probably the most important thing in the evaluation of quantifiable forensic evidence. For a brief account of the use of Likelihood Ratios in a real-world $150 million telephone fraud case click here. A fuller account of the case is here. A report in another real-world forensic voice comparison case involving (non-terrorist) car-bombing is here. For a very much more technical account of how Likelihood Ratio-based forensic voice comparison can emulate DNA, click here. For a presentation on the assessment of strength of evidence in voice lineups click here. To find out more about the proper evaluation of forensic evidence with Likelihood Ratios you should visit Geoff Morrison's web-page.

Most of my current forensic voice comparison research deals with testing the strength of evidence obtainable from the kind of speech features to be found in real-world case-work. My papers on forensic voice comparison, and related subjects like the use of Bayes' theorem in historical linguistics, can be viewed and downloaded here. They include work on Japanese and Chinese as well as Australian English.


I consult as a forensic speech scientist for the police and private clients, and I have been doing forensic voice comparison, disputed utterance and trademark case-work on Australian English and varieties of Chinese for nearly 20 years. To read a disputed utterance report in a real murder case click here.

To make initial inquiries about retaining my services, please use this email link, giving the particulars of the case (no sound files please!). We can then talk over the phone to see to what extent my services can be of use.


I cut my descriptive phonetic teeth on the Chinese Wu dialects in the 1970's, writing my field-work PhD thesis on the acoustic phonetics of Zhenhai 鎮海, a dialect south of Shanghai, near Ningbo 甯波. Wu dialects have some of the most complex tones and tone-sandhi systems in the world, and I was hooked on their phonological complexites from the start (much more interesting than forensic voice comparison!).

For the last 20 years I have been analysing data from an extensive compliation of recordings of Zhejiang Wu dialects made, by colleagues and myself, mostly in the 70's 80's and 90's. Some of the recordings are now quite old, and of speech patterns that are disappearing or have disappeared. I have now started to make available some of the more interesting Wu data on web-pages here.

Historical recordings, but of a very different nature, can be found here: first-hand accounts of three world-war II episodes involving the elite German Brandenburger commando unit recounted by a person who actually took part.

Apart from Wu dialects, I have also published on tones in Cantonese and Thai. My papers on tonal phonetics and related topics like Tibetan Spelling Chant can be viewed and downloaded here.


About to step the Sussex Constabulary full-back. In such a moment one is on autopilot and I recall absolutely nothing of it, although I do remember getting my ear bitten sometime in the match.

Although I of course didn't realise it at the time, I was immensely lucky to benefit from the no-frills thoroughness of a typical 1960's English Grammar School education. And fantastic teachers inspired perhaps by some kind of post-war idealism. Too late to thank them now, but I guess they would have already known about Huxley's purpose of education being to make your mind a pleasant place to spend your life in.

You were forced to specialise quite early, which usually meant deciding between science and arts. I really enjoyed maths without ever being in the slightest bit good at it - the harder I tried, the wronger the answer - and I kept in touch with science by assembling skeletons for the school's biology lab, often from donations from the local butcher, road-kill or deceased pets (which died of natural causes I hasten to add). Osteology taught me about STRUCTURE. My real love though was sport, but you couldn't take exams in Rugby and Cricket and although being a hopeless speller I did French German Latin and Spanish instead.

In a working year between school and university I started to teach myself Chinese and Japanese, after which I read German and Russian for my honours undergraduate degree at Manchester University. German at Manchester provided a fantastically thorough grounding, engagingly lectured. It covered not only German literature and history but, because the then department head Prof. Keller was a linguist, as many linguistic aspects of German as you could poke a stick at (Gothic, German Dialects, the Orwellian language manipulation of the Third Reich, you name it). My degree also luckily included a year and a half's study of Chinese and Japanese at Erlangen University in (then) West Germany, interspersed with labouring work on various building sites around Nuremberg. I followed that with an MA in Linguistics, where I discovered I was a phonetician. I wrote my MA thesis on the acoustic-phonetic description of tones in a Chinese Wu dialect, also discovering in the process that speakers of tone languages can get up to vastly more interesting things with their vocal cords than I had learnt from Mandarin.

Giving a speech, in conservative Zhenhai dialect and uncustomary mid 70's sartorial elegance, at the Hong Kong Ningpo Residents' Association Spring Festival dinner. The conspicuous lip-protrusion is from the morpheme 水 [sʷʮ] 'water' in the compound word 酒水'banquet' ['ʨɪʊsʷ 45]. [sʷʮ] loses its vowel and tonal pitch due to aerodynamic reasons.

I spent two years doing (not very uncomfortable) field-work with old Wu dialect speakers in Hong Kong and Taiwan for my PhD at Cambridge - also on Chinese phonetics. After that I left England for a teaching job in much warmer weather, thus buying about ten years before the scourge of economic rationalsim started to destroy higher education Downunder. I learnt heaps there - in the then heavily empirical-descriptive Linguistics Department at the Australian National University - where I taught Phonetics, Chinese Linguistics, Phonology and Introduction to Linguistics for over 30 years. The beautiful zero-radiograph of a lowered-larynx [u] vowel at the top of this page, its mouthful of tooth fillings attesting to my flouride-deprived childhood, is from an early study on the relationship between velum and vowel height in Cardinal Vowels done to help the cranio-facial reconstruction unit of the (then) Adelaide Women's and Children's Hospital. I have always found pictures of the real thing very useful in teaching phonetics.

Sometime in the mid 80's one of my ex-students launched my forensic speech science activities by dobbing me in to a Melbourne solicitor, and my research expanded, quite naturally really, from what speakers of tone languages have in common, into how individual speakers differ.

In 2004 I was a British Academy Visiting Professor at Edinburgh University's Joseph Bell Centre for Forensic Statistics and Legal Reasoning. Looking back, this really was a game-changer. At that time, one of the obvious problems with estimating likelihood ratios from speech acoustics was that the variables were always correlated and this meant that they could not be cavalierly combined  àla naive Bayes. The forensic statistic group at Edinburgh had developed a method of taking correlation into account when estimating likelihood ratios and I went there to learn from Prof. Colin Aitken how to apply it with their Multivariate Likelihood Ratio. Dr David Lucy took me under his wing and patiently introduced me to R in which it was all coded. Their multivariate likelihood ratio is still used in forensic voice comparison research and case-work: forensic voice comparison has a lot to thank them for, as do I.

Extravagantly well-hung male fox prowling across a piano keyboard. Notable is not only the hyoid bone posterior to the lower mandible (every self-respecting mamalian skeleton should have one) but also a penis bone, my placement of which now looks way too low.

I became chairman of the Forensic Speech Science Committee of the Australasian Speech Science and Technology Association, and subsequently a member of the Australian Academy of Forensic Sciences. I have also held positions on the Council of the International Phonetics Association and the research committee of the National Voice Centre of Australia (now long defunct - another victim of our incessant funding cuts).

In 2012 I was a Visiting Professor at the Hong Kong University of Science and Technology, where I taught Chinese Phonetics syllabus here, and Forensic Voice Comparison in Cantonese syllabus here. (The forensic course was intended to see whether the Likelihood Ratio-based approach to forensic speaker recognition was too difficult to learn. Apparently not: the seven papers resulting from the course can be found in my forensic publications under years 2012 and 2016.)


Late in 2012 I was given a superbly produced and fascinating book called Quantitative Approaches to Problems in Linguistics. Ah! but this was a very, very special book: as well as a beautifully caligraphed Chinese poem, it also had my name on the cover! A festschrift, nothing less, from former students and colleagues. I continue to be overwhelmed by this gesture, even tearing-up when I look at it (but not to the extent that it prevents me from leaving said book around in plain view). The singularly appropriate poem on the cover reads: I left my home when young; old, I have now returned. My native accent hasn't changed, but my hair is grey. The little children don't recognise me. They laugh and ask "what place are you from, stranger?" [少小離家老大回/鄉言無改鬢毛衰/兒童相見不相識/笑問客從何處來].

In 2015, I gave a week-long workshop on 'Tonetics' (tonal phonetics) at the University of Hong Kong. The schedule is here. In 2018 I was Visiting Professor at China's Southwest University of Political Science and Law where I gave a month's introduction to Bayes' Theorem and its application to forensic voice comparison. In May 2019 I gave a week-long intensive course in Tonetics at the Chinese University of Hong Kong. I gave a keynote presentation The Best of Tones, the Worst of Tones at the Australasian Speech Science & Technology Conference in Canberra in December 2022. You will find the abstract here.

recording Wenzhou in 2017 Checking the tone sandhi in some polysyllabic words in Wenzhou dialect with a very patientinformant in 2017.

At the moment I am busy juggling research into tones, research into forensic voice comparison, and real-world forensic case-work. And the occasional skeleton. I am still not very good at maths, but have learnt not to concentrate too hard and consequently now get more correct answers. I have resigned myself to never being able to spell.