This book, by an enviably trained phonetician with an enviable classical background to boot, is the most recent, complete and accurate description available of what classical Latin sounded like. Though I imagine a reader not trained in phonetics and phonology might have a hard time following it, this obstacle is mitigated by the inclusion of a glossary explaining technical terminology.
I do take issues with certain things though.
It seems a shame to me that the loss of distinction between T and D in word-final position was not mentioned. (c.f. Quintilian "ut 'ad' cum esset præpositio, 'd' litteram, cum autem coniuctio, 't' acciperet.")
On the issue of short I after QV, Allen leaves the reader with a somewhat confusing impression. On the one hand, he relies on grammarians such as Priscian to show that the I in such words as "quis" and "vir" had a more rounded quality. On the other, he quotes Priscian to the effect that the QV before front vowels (like "I") had a fronted secondary articulation. One is left wondering whether to front the labial consonant, round the front vowel, or both in words like "quis." Some commentary on this point would have been in order.
On two points, Allen makes some assumptions which, to me, seem unjustified:
In a section on the voiceless plosives P, T, K on pages 12-13, Allen is quite correct in advising that English-speakers not attempt to replace their aspirated stops with unaspirated ones. But to assume that Latin voiceless stops may have had aspirated allophones is quite another matter. It is, naturally, impossible to deny that they did, but all available evidence indicates that they did not (at least, not in the mouth of a native speaker) until the second century (where they were marked with the familiar digraphs ch, th, ph already known from greek borrowings.) Allen's historical evidence (based largely on the voicing of early borrowings such as "Burrus" for greek "Pyrrhus") is not sufficient to indicate otherwise. Most of the borrowings mentioned are not in fact Greek borrowings but non-indo-european words borrowed independently into Greek and Latin from a third source. This third source may have had voiced and voiceless stops in different dialects, which is a much more plausible explanation.
In the section on word-final M on pages 30-31, Allen mentions the assumed phenomenon of nasalization. However, the fact that word-final M is not counted in scansion does have *other* plausible explanations, most notably the theory that it represented a voiceless bilabial approximant. The lack of word-final F coupled with the metrical behavior of H makes this theory, at the very least, worth mentioning.
Other than the above shortcomings, it is a beautifully researched book written in a kind of lucid prose which, ironically enough, is rare among modern linguists.