Until recently, we had little idea how ancient Greek music might have sounded. But now, for the first time in 2,500 years, we have a chance of actually putting together the available information. Our notion of what the music of ancient Greece sounded like comes from fragments of 60 musical scores on Egyptian papyri, the first of which was discovered in the 19th century. Naturally, it is unlike anything imagined by the inventors of opera, who in Renaissance Italy decided to create a new kind of music drama in an attempt to re-create ancient Greek theatre. The Renaissance scholars' biggest problem was that they hadn't the foggiest idea of how that theatre looked or sounded: notably, was it spoken or sung?

I was interviewing Oliver Taplin, professor of classics at Magdalen College, for a Radio 4 programme when the idea of reconstructing the music of ancient Greece popped into my head. The programme, In Chorus, considers the history and attractions of an activity - singing together - that seems to have existed for as long as society has. Its origins predate the great Greek dramas of the 5th century BC, which emerged from even earlier Dionysiac (read: drunken) celebrations; but its first documented appearance is in those tragedies of Aeschylus, Sophocles and Euripides, where the Chorus, while not taking part in the action of the play, comments upon it and reacts to it.

Taplin was able to answer that first question for me. At the time of the Renaissance, it was believed all the acting parts in Greek theatre had been sung. "We now believe only the choruses were," he says. "But they did have the right idea about it being one-note-per-syllable: audibility was the crucial thing. These Greek choruses were highly trained - and entirely paid for by rich citizens as a form of taxation." (Bleed the rich to pay for a Covent Garden chorus? Not a bad idea.)

"Greek tragedy is an odd form of drama because the plot keeps getting interrupted by this chorus, who keep insisting on singing and dancing," Taplin continues. It sounds more like a musical than an opera, "except the choruses weren't part of the plot; more a kind of meditation on it, like the chorales in a Bach Passion".

I was nagged by a desire to hear what this music - such an integral part of the drama - sounded like. I asked Taplin if we could use his translation of a chorus from Sophocles' Oedipus in the programme. He agreed, but only if I got it set to music. I decided to make something as near as possible to the spirit of the original - and to record and broadcast it, too.

Taplin pointed me towards Roxanna Panufnik, who in 2003 had been commissioned by the English National Ballet to compose a ballet score on the Greek myth of the swan-fancying Leda. (Sadly, the ENB ran out of money and the ballet has yet to be performed.) Panufnik had gone to the greatest authority on ancient Greek music, Dr Martin L West of All Souls, Oxford, for clues about how to compose it. "I wanted to understand the structures of Greek dance, the equivalent of our minuets and so on, and to understand their principles of melody, rhythm and modes, which were quite different from ours," says Panufnik.

What she found was surprising. "It was pretty funky stuff, with a kind of rhythmic wildness that didn't appear again in western music until the beginning of the 20th century." West, amazingly, had managed to reconstruct - from those 60 fragments and the contemporary descriptions - an entire system of modes and rhythms. "There were ancient technical writings about how to sing the music and how to interpret the notation, but nobody could put them together," he says. "People have played and recorded the fragments of scores, but as far as I know nobody has tried to compose in the style of ancient Greece - although I did once put a scratch choir together for a demonstration at an Oxford triennial conference."

The sequence I asked Panufnik to set to music is Taplin's translation from the final chorus of the Sophocles play, when Oedipus has had to admit the truth - that he has unwittingly killed his father and married his mother. The tragic hero vows never to look on the sun's light again, and the chorus of old men - Oedipus's companions - try to draw some desperate moral from the situation. "It's a lament, so I used the Dorian mode - different from the Dorian mode we know from Gregorian plainchant in that it has extra quarter-tones above the second and sixth, and the eighth is in fact the ninth ..." explains Panufnik (in other words, take the white notes on a piano from D to D, insert extra notes between E and F and B and C, miss out the top D and play E instead.)

In a sense, Greek chant is not unlike Gregorian chant. The chorus (of 12 or 15 men) sang in unison, accompanied by a double-reed instrument called an aulete - and by all accounts, danced at the same time, albeit a stately, tai-chi-style dancing of the sort you might see in a Robert Wilson production. "It's hard to say whether different composers had different styles," says Taplin. "The rhythm of the music follows the rhythm of the words, so in a sense there's less freedom." Panufnik agrees: "The words drive you forward. The biggest difficulty for me was that I couldn't modulate to another key, which we are sort of conditioned to do and which gives music a shape and direction. Here that is all provided by the words, the drama."

The other main element is the rhythm. Panufnik used a dochmiac rhythm (a foot of eight uneven syllables in pairs of five and three) and uneven bars of 5/8, 7/8, 6/8 and 8/8 to express the emotional anguish of the passage (The all-seeing eye of time/ has - despite you - hunted you, and discovered the same womb/ gave birth to your children too; judged your marriage as ... incestuous/ wretched son of Laios! All I can do is lament you/ pouring sorrow from my mouth ...)

Hearing this sung for the first time was eerie: voices from the past if ever there were. Panufnik's setting captures the strangeness of this archaic, alien music - its quarter tones, its rhythmic shifts, its visceral rawness. Like much else of ancient Greece, this music seems to have died completely, though there may be hints in south-east European folk music, and notably in the "open-throat" singing technique of Bulgarian folk choirs. It was easy to believe that the origins of this music lay in Dionysiac rites, where the chorus of lithe young men would prance about in leather shorts adorned with a huge phallus.

The choir who recorded Panufnik's piece for us, Canticum, put aside ecclesiastical leanings to give us full-blooded Balkan emoting, turning a Pimlico church into the theatre of Epidauros - though alas, we couldn't convince them to don their strap-ons.