Romantic Distances

Some records are made to document a concert, others to gather a repertoire, still others to meet the demands of the record market. Corde di Luna, released by Harmonia Mundi, belongs to a rarer category: that of projects conceived as a single artistic idea, matured over time and destined to find in the recording not its starting point but a destination. Giulia Semenzato produced it herself; the session took place in July 2024 at the Operahuset in Oslo — a theatre, not a studio — and the record is released two years later, on 17 July 2026.

Those two years are not an editorial detail, and the genesis of the project can now be reconstructed with precision. On 1 August 2022, at the Festival della Valle d’Itria in Martina Franca, Semenzato and Duncumb performed together in the “Il canto degli ulivi” series with a programme already built around Sor, Giuliani, Doisy and Schubert, but still substantially different from what would become Corde di Luna: it included pieces that never made it onto the record, such as Sor’s Capriccio op. 50 “Le calme,” Schubert’s Liebesbotschaft, and a Tarantella by Mertz different from La Danza. A year and a half later, on 8 April 2024, the duo performed at the Théâtre de l’Athénée in Paris, in the “Les Lundis Musicaux” series: and here the programme coincides, almost piece for piece, with what would be recorded in Oslo three months later — Giuliani’s six Cavatine, Doisy’s three Romances, Mertz’s Liebeslied, Schubert’s Gretchen and Ständchen, Sor’s Fantaisie op. 30 and Seguidillas, Rossini to close. Corde di Luna, in short, does not capture a single recital: it captures the outcome of work that lasted at least two years, in which a broader programme was progressively refined before audiences until it became, already in Paris, precisely the sequence that would later be recorded.

The project, then, is well conceived, and it is worth saying so before any reservation. At its centre are not Rossini and Schubert, as the preview released these past weeks might suggest, but Sor and Giuliani: the two composers who wrote for the guitar knowing exactly what it was, and who alone account for two-thirds of the programme’s fifty-eight minutes. Schubert himself does not appear in piano reduction, but in the transcriptions of two guitarists — Mertz for the Ständchen, Napoléon Coste for Gretchen am Spinnrade. And Duncumb plays a S. N. Gade guitar, built in Copenhagen in 1828: the same year Sor published in Paris the Fantaisie et Variations brillantes op. 30 that appears on the track list. Whoever built this programme knows precisely what they are doing, and the chosen instrument is not a liner-note detail: it is part of the statement of intent.

The promise, too, is clear, provided it is read in the right place. Harmonia Mundi’s social media presentation speaks of a loving dialogue between voice and guitar, “much more than a recital,” of a journey through human passions. In the booklet, however, the word belongs to Jadran Duncumb, and it says something else. Not dialogic parity, but an ancient bond between the human voice and the plucked string — an association, he writes, that goes back to when people first began writing about music: from Orpheus to the French airs de cour, from serenades under balconies to Joni Mitchell. The programme is “Romantic,” he concludes, but the romanticism found in it is timeless. It is this second promise, not the first, that truly governs the record, and the arithmetic confirms it: of fifty-eight minutes, nearly sixteen have no voice at all — the Fantaisie op. 30 and Mertz’s Liebeslied, well over a quarter of the album. Anyone looking here for a dialogue between equals, in the sense the label promises, is looking for something the record itself never intended to offer.

There is then a third promise, one the record writes down nowhere and yet keeps from the first minute: closeness. It is heard from near, in an almost physical sense. In the Liebeslied you can hear Duncumb’s hand running over the strings; in the Ständchen you can hear Semenzato drawing breath before the phrase. These are not recording flaws: they are the choice, and the booklet grounds it fully without ever naming it as such. Matteo Carcassi was a favourite of the Parisian salons in the era when Rossini lived there; Mertz’s Bardenklänge are modelled on the domestic piano miniatures of Schumann and Mendelssohn; Schubert’s Lieder were often published with a guitar accompaniment precisely to reach the rising middle class, who devoted their evenings in the salon to that music. The historical listening position for this repertoire is not the fifteenth row of a theatre: it is the armchair two metres from the performers. And at two metres the hand is heard, and the breath too. To ask that these traces be scrubbed away would be to ask for music stripped of its own body — exactly what this repertoire has never been, nor wishes to be here.

Yet on this record, so well conceived, two distinct flaws coexist, and it is worth telling them apart from here on, because they arise from different causes and do not explain one another.

The first concerns that very closeness: it was chosen once, and never reconsidered. Because at two metres from a guitar you hear everything, and here the guitar, more than once, is not heard. In La Danza it is the tambourine that covers it, and the choice belongs to the performers, not to Carcassi: the booklet notes that Carcassi, in transcribing Rossini’s Soirées musicales, deliberately avoided imitating the piano and instead kept to the folk roots of the music and the instrument, and that it is in that spirit that Maxime Pidoux — already principal timpanist at La Scala, not a studio stand-in — joins them on the tambourine. The choice is consistent with the transcription, and is not in question in itself. But if the merit of that transcription lies precisely in having entrusted the guitar with the tarantella’s popular character, then the guitar figuration is the piece’s rhythmic engine, and the percussion brought in to underline it ends up, in the final mix, covering the very instrument that produces that character. Doubling is one thing; unintentionally replacing is another.

In Sor’s Seguidillas there is no tambourine, and the guitar recedes just the same — here with a mitigating factor, since these are salon songs in which the voice carries and the guitar accompanies: there is no dialogue to sacrifice in that writing, because none was intended. That this is a matter of levels, though, and not of listening perspective, is shown by Mertz’s Liebeslied, the only solo-guitar page in the opening minutes: there the instrument is alone, and you hear it right down into the fingers of whoever plays it. The rustle of the hand on the strings, audible and persistent, tells us the microphone is close. It is not, then, physical distance that robs the guitar of definition in the songs, because that distance is, in all likelihood, the same. It is the level to which the channel is brought in the mixing stage, and it is that level that changes from one piece to the next.

That the record can also succeed, and succeed well, is shown by the two Schubert pages, perhaps the most convincing of the entire listening. In the Ständchen the balance is there, and the result is the most beautiful thing encountered in these tracks: the serenade does exactly what a serenade should do, which is very nearly make the listener fall in love. The breath is handled naturally, and if anything the plea has an urgency that does not suit it — whoever pleads beneath a window, after all, has the whole night ahead. In Gretchen am Spinnrade the spinning wheel truly turns: Coste does not simply copy the piano part, but builds for the guitar its own figuration of the turning wheel, imitating — as Schubert’s original piano writing does — its motion and its pedals. There the instrument has something precise to defend, and it defends it. In the moments when it can speak in full — these two, above all others — the guitar asserts itself without visible effort, and it is precisely there that the regret for the rest of the record grows sharper: not for any lack of quality in the performance, but for the way that quality is distributed across the programme.

The clearest proof of the flaw, however, lies elsewhere: in the romance that closes François Doisy’s Trois Romances, titled À ma guitare. It is the only page on the entire record that speaks, literally, of the guitar: ma guitare, oh! la seule amie qui puisse adoucir mes malheurs… viens mêler tes chants à mes pleurs, come mingle your songs with my tears, the text sings, addressing the instrument as a confidant. In those songs, here, one struggles to enter: the guitar all but vanishes during the singing, and resurfaces the instant the voice falls silent, only to disappear again as soon as the verse resumes. This is not a matter of subtle writing, hard to catch: the musical material is there in full, and it is heard clearly every time the voice leaves room. The guitar’s level has simply been set just below the soprano’s, and when she sings it ends up beneath it — in precisely the piece where, given the words it carries, it deserved to stay afloat.

This first flaw, then, does not arise from the conception, which remains excellent, nor from the performance, attentive and intelligent on every page. It arises from the fact that closeness — the record’s finest quality, the one that makes it feel so alive to listen to — was set once and never reconsidered afterward: the same distance for a tarantella accompanied by tambourine, for three salon seguidillas, for an eighteenth-century romance written for voice and lute alone, each with different needs.

The second flaw is of a different nature, and — paradoxically — it is revealed by the technically best-executed tracks: Giuliani’s Sei Cavatine op. 39. Here the balance is the best on the whole record: the guitar is heard, has room, has nothing to defend itself from, and none of the objections raised so far apply. And it is precisely here, with the balance fully resolved, that the listening struggles to leave a mark. The booklet argues that in Giuliani, as in the Baroque repertoire Semenzato and Duncumb say lies behind them, musical ideas are never imposed on the poetic text: they spring from the words of Metastasio and of anonymous authors, and paint them through melody, articulation, harmony, shifting — to use the booklet’s own phrase — from one phrase to the next. The six cavatinas do indeed set opposite registers side by side: the breathless joy of Par che di giubilo l’alma deliri and the weeping that begs pity from a heart of stone in Alle mie tante lagrime. Heard repeatedly, hours apart, the voice that sings them remains, however, substantially the same: watchful, composed, identical to itself from one piece to the next. The restraint that served La Danza so well does not, here, bend to the contrast that the booklet itself claims as this part of the programme’s guiding principle.

This time it is not a matter of mixing, and there is no rustle of a hand to lean the observation on: it is that, faced with texts of opposite sign, the vocal line does not substantially change colour. It is the hardest flaw to put into words, because it has no external technical culprit to name, and concerns an interpretive choice directly. It does, however, have a proof as repeatable as the others, and anyone can carry it out in a few minutes: it takes only two cavatinas heard side by side, with the single question of whether anything, in the colour of the voice, truly moves between joy and tears.

It is worth noting, in passing, that the second half of the record holds up better than the first — not decisively, but perceptibly: it is there that both Giuliani’s Cavatine and the Schubert pages are concentrated, the repertoire in which the guitar has more written space and therefore more chances to assert itself even at an unchanged level. This is not enough to correct the first flaw, which cuts across the whole programme, but it does indirectly confirm that when the writing helps the guitar, the record responds.

These, then, are two distinct flaws, not one wearing two disguises. The first strikes wherever the writing calls for reciprocity between the instruments and no one in the room had the explicit task of correcting the level case by case. The second persists even where the sound is perfectly right, and belongs, therefore, to the performer, not to the sound engineering. The record is not simply weaker than expected: it has one flaw that could be corrected by resetting a fader, and another that could be corrected only by changing one’s mind about a musical phrase. The first is almost an accident along the way, however recurrent; the second says something more about the approach with which this repertoire — so well chosen, and so long seasoned on stage — was handled at the interpretive level.

This writer will not return to this record often — not for any insufficiency, but because two small limitations, added up over a full hour of listening, weigh more than their arithmetical sum.

One small clarification, to close the review. This writer published, in early July 2026, La forma della voce. Etica e tecnica del canto, an essay dedicated precisely to Giulia Semenzato. Let the reader bear this in mind: the pages above do not arise from distance, but from esteem. And it is precisely for this reason that they do not stay silent about what fails to convince.