HAD I BEEN ASKED, I would not have put a new Tosca%N on a list of the 50 recordings most needed at the moment. But Philips has gone ahead and done it anyway, and the result on Philips 6700 108 (two records), with Colin Davis conducting and Montserrat Caballe, Jose Carreras and Ingvar Wixell in the leading roles, is a complete triumph. One may prefer an isolated scene in some other recording (Callas at the death of Scarpia, for example), but this is a very carefully integrated performance, strong in dramatic and musically unbeatable. Davis conducts it as a single, prolonged music-drama, not an interrupted series of big moments, and the result is impressive.

Equally impressive are the voices. Caballe is perhaps a shade sweet and gentle for the role as it has evolved in modern interpretations, but her Tosca is dramatically believable and vocally superb. Carreras (except for rare moments of self-indulgence) is that rare phenemenon, a tenor more intent on total musical effect than his share of the applause. Wixell's Scarpia is a shade understated dramatically (we can use a bit of that), but tonally superb. The sound is rich and detailed, the Covent Garden Orchestra and Chorus flawless.

In sum, Philips has tackled one of the most popular and often-recorded operas in the repertoire and improved on the competition in significant ways. If recording companies have a responsibility to the art of music (and I think they do), this is one way it can be fulfilfed while recording music which has already been over-recorded. Philips has, of course, earned the right to an occasional Tosca - for example, with the series of Haydon operas it is recording under Antal Dorati's direction.

Recent months, in fact, have been good for those opera fans who want to hear music outside the basic repertoire. From Columbia, there have been solid if not definitive first recordings of works as diverse as Meyerbeer's Le Prophete (M4 343-40, four records), the grand-daddy of modern grand opera and Gustave Charpentier's Louise (M3 34207, three records), which was, we are told, the first French opera to give a sympathetic, understanding portrayal of working-class people, their personal problems and conflicts. Fans of Delius (a small but growing group, to which I do not belong) have been well served by Angel with good recordings of his Fennimore and Gerda (SX 3835, two records) and Koanga (S 3808X, two records). Having played each of the above once and felt no great desire to do so immediately again, I hesitate to recommend to anyone else the considerable expense involved in purchasing these sets. But it is largely a matter of personal taste, and I trust that many readers will rejoice at the availability of one or another of these items.

Listed and briefly annotated below are several recordings of seldom-heard operas that I have felt inclined to play more than once in recent weeks:

DONIZETTI: Gemma di Vergy. Montserrat Caballe, Paul Plishka, Louis Quilico; Schola Cantorum; Opera Orchestra (Columbia M3 34575, three records). The libretto, full of thwarted love and vengeful bloodshed, is stronger than some used by Donizetti, which doesn't mean it has to be very strong; the music is what one expects from him - rather crude and four-square in the choral and orchestral writing, much better for the solo singers and superb when the soprano comes in. With an adequate supporting cast and good conducting, it is primarily a showcase for Caballe, for whom the opera has been revived after a long absence from the repertoire, and I think a single disc of high lights (devoted largely to the excellent finale) might have served most people's needs. The recording was made at a concert performance; there are a few coughs and some musical infelicities that would have been edited out of a studio recording, but nothing seriously disturbing.

RACHMANINOFF: Francesca da Rimini, Vladimir Atlantov, Makvala Kasrashvili, Mikhail Maslov, Alexander Laptev; Bloshoi Orchestra and Chorus, Mark Ermler conductor (Columbia-/Melodiya M2 34577, two records). Based on the same episode from Dante that supplied Tchaikovsky with the subject of one of his best tone poems, this opera's two brief acts are fairly standard romantic melodrama, though well made. What gives it special distinction is its long, atmospheric prologue, in which Dante, being led through Hell by the shade of Virgil, encounters the unfortunate lovers and asks to hear their story; the music portrays the scene vividly. Tenor Vladimir Atlantov is outstanding in the role of Paolo, and the whole production is well cast and conducted. On side 4, Atlantov has a recital of Russian arias, a very substanial bonus for fans of this extraordinary singer. The libretto includes a Russian text (in our alphabet) for the arias as well as the opera.

HANDEL: Semele. Sheila Armstrong, Helen Watts, Robert Tear, Justino Diaz; Amor Artis Chorale: English Chamber Orchestra, Johannes Somary conductor (Vanguard VCS 10127/8/9, three records). Handel's claim to recognition as one of the world's great opera composers is beginning to get widespread and serious attention, and this superb composition helps to explain why. Its story and conventions differ drastically from those of the more familiar operas of Mozart and the romantics, but this story of Zeus in love is just as credible as The Magic Flute or most of Verdi and the music is glorious - one of the high points of baroque art. The voices here are a bit below the international superstar level but nonetheless excellent, and the interpretation is beautifully styled. Vanguard is reissuing this set at half-price, a strong inducement for music-lovers in search of something different.

JACOPO PERI: Euridice Nerina Santini, Rodolfo Farolfi; Coro Polifonico di Milano; Solisti di Milano, Angel Ephrikian conductor (Telefunken SAWT 9603/4-B, two records). Dating from 1600 (a good round-number year for a major historic development), this is the first serious opera in music history; earlier staged works in the form of a series of madrigals don't really qualify, and neither does Dafne , a comic pastoral by the same composer that dates from a few years earlier. Its simple, predominantly stately but flexibly expressive cadences mark one of the most significant revolutions in music. Peri falls slightly short of the emotional depths reached by Monteverdi (who began writing in a similar style a few years later), but listeners will be reminded of that great composer frequently in the simple, flexibly expressive melodic contours of this music. I find myself drawn to this work again and again, although the voices in this recording are serviceable rather than dazzling, and the reason is not merely the music's historic importance. Peri has taken one of the great themes in opera - one that served later composers as diverse as Gluck and Offenbach - and given it a thoroughly worthy treatment, one in which the music constantly heightens the effect of the words and adds an extra dimension to the story. Peri not only created a new kind of music, a new kind of drama; he struck the right style and tone the first time with remarkable precision, laying a foundation on which a rich art form was able to grow and is still growing. Those who are really serious about opera should know this music.