The Golberg Project

Mathias Halvorsen


The Goldberg Project consisted of three concerts with Bach’s Goldberg Variations performed in Berlin, Oslo and Drøbak in the summer of 2015. From each concert a recording was made that highlighted a different aspect of classical music recording. The project was a collaboration between pianist Mathias Halvorsen and sound engineer Johann Günther and supported by Kulturrådet and Frogn kommune.


The Goldberg Project was intended as an inquiry into the aesthetic questions surrounding the current practice of making classical music recordings and their relationship to live performance. As a classical performer heavily steeped in the tradition I wanted to confront my instincts, adopted and cultivated since childhood. The goal was to approach the process of recording in a different way; rejecting the ideal of highly formalised and perfected output that currently dominates the field. The process was structured around a planned set of unedited recordings, executed in different spaces on instruments with strong characteristics.

Early on I joined forces with sound engineer Johann Günther and together we set out to find locations and instruments that could prove interesting for the project. We wanted to create an atmosphere in the recordings that would be in active dialogue with the instrument at hand, letting the surrounding circumstances colour each performance. The choice not to exclude or edit out mistakes and other spontaneous events was made from the start, thus bridging the gap between recording- and performance practice. The decision to do three full recordings of the same piece was made to test the ideas of the project against the perception that recordings of standard repertoire often sound very similar. Three separate recordings would also allow us to isolate and explore different aspects of the project in detail. We chose the Bach’s Goldberg Variations as a subject because it has certain traits that would allow us to communicate the ideas of the project in a clear way. The piece is well known, has a transparent structure and a narrative arch that would suit our approach rather well.

Several artists have engaged with ideas addressed in this project. The album Wagner e Venezia by Uri Caine (Winter&Winter, 1997), recorded live from an outside café in the middle of Venice, was an early inspiration for further inquiry. I think it is a great example of how the sound of the environment can impact how the listener experiences the music. Andràs Schiff’s recording of Beethoven’s Diabelli Variations (ECM, 2013) on two different pianos is also an interesting study on how the instrument can influence an interpretation.

Part 1 - Berlin

With the first recording we wanted to explore the conflict between a player’s intentions and the restrictions imposed by the instrument at hand. The location was a small concert studio in Berlin. Johann tracked down the place, which suited us especially well because of the instrument: an old Steinweg-Welte grand piano. This special and rare kind of instrument was equipped with a Welte-mechanism, which made it possible for the instrument to reproduce recorded performances all by itself. Through an elaborate system of bellows moved by a changeable air supply controlled by a roll of perforated paper, the piano was able to reproduce not only the pitch but also the volume of each individual key.

As a self-playing grand piano, the instrument had several unique attributes we found instantly likeable. Firstly, the sound of the instrument was very special. It had an old, mellow and slightly percussive sound that is very hard to find in newer instruments. Secondly, to make room for the Welte-mechanism the piece of wood extending from the keys to the hammers was more than twice as long as that of a modern grand piano. This made it physically a very heavy instrument to play, seriously constraining what you could and could not do musically. Thus, we were confronted with an intriguing fact: the physical nature of the instrument was forcing the direction of the interpretation and influencing tempo, timings and voicing. Adapting to the playability of the instrument at hand each time is something present in every interpretation, but this technical variable is rarely highlighted in a performance or recording.[1]

We often view a musical performance as something purely and fully formed in the mind of the musician before the concert starts. Other factors influencing the final output are often seen as disturbances to be smoothed over, corrected, or even eliminated. The Steinweg-Welte gave us the opportunity to make a recording where one can hear the performer trying to adjust to a very difficult instrument. The listener can hear me succeed, struggle and fail (sometimes painfully badly, sometimes strangely beautifully and sometimes both at the same time) while working my way through Bach's massive set of variations. What results is an interpretation torn between the pre-planned intentions for the piece and the struggles, often unexpected, with the technical challenges posed by the instrument. In the end the player’s spontaneous reaction patterns are exposed with a rarely heard clarity. In addition, Johann’s choices with the sound design highlight these aspects through a close set-up, producing a clear and transparent sound.


Example 1: Recording from the Berlin performance

Part 2 - Drøbak

The subject of the first recording had been an old piano in a small studio in Berlin. For the second we choose a new Yamaha grand piano situated in a beautiful wooden church in Drøbak, a small seaside town in the South of Norway. The piano had a dry, crisp sound with a mechanism of keys and hammers in good working order. The performance was advertised in the local newspaper and was more of a fully-fledged concert than what had taken place at the Berlin venue (which was a very intimate affair). The biggest change, however, ended up being our spontaneous decision to do a pre-recording of the entire piece, giving us the opportunity to edit the concert recording afterwards. This is a standard approach to classical live recordings and even though not a part of our original plan, we decided to leave the choice open whether to use the added material or not.

It took us a full year to return to the recording from Drøbak. Listening back to it after having completed the two unedited instalments of the project (Berlin and Oslo versions), we were inspired to take this recording in an unexpected direction. The search for expression in uncontrolled events undergone in the other two recordings had changed our perception of mistakes and other impurities. Usually when encountering a mistake in the editing process, the take in question (or at least the surrounding few seconds) are cut out or excluded by default. This time, however, we approached such moments with enthusiasm, choosing whether to include it or not, based on the perceived quality of the accident in question. We were now approaching the editing process not only with the traditional concerns to exclude mistakes, but also to edit them in when we liked what we heard. This appreciation of the quality of impurity in a classical editing process signalled a radical shift in our perception of musical quality.

Some mistakes had characteristics that where easier to enjoy and more accessibly expressive than others. They seemed to provide a rare sense of intimacy with the performance, perhaps giving the listener a glimpse behind the well-rehearsed, professional front presented by most performers. During a concert, I can think of few things more personal and engaging than a mistake, whichever form it might take. It is also both exciting and instructive to hear the subsequent reaction of the exposed artist as the performance continues to progress. The obvious emotional echoes of a performer trying to either divert attention from a mistake or working hard to stabilise a shaky passage, are highlighted and work as a storyline on their own, alongside and interwoven with the progression of Bach’s piece. By experiencing the performer’s momentary lack of control, the listener might become more perceptive to the subtler musical changes occurring as a result of the psychological reactions of the player.

The most inspiring and thrilling accidents occurred when I would forget the memorised score — but still knew where I should eventually end up. Listening back, every second of the resulting improvisation brought noticeable tension to the performance. Moments like this put rapid instinctive musical responses front and centre, which we perceived as an obvious diversion from the intended interpretation. I would for example miss the starting note of a melody, having to improvise a musical line fitting with the remembered chords while trying to find a natural transition back the score. When I would find my way back and the piece would move on, the effect of the mistake on the performance often lingered on long after the return to safety. When this happened during one of the many repeats in the Goldberg Variations, you would end up with a kind of naturally occurring organic variation, very different than the more common addition of trills and slight melodic discrepancies. When working our way through the concert recording and the pre-takes, we ended up not only editing out accidents from the concert but just as often adding them to the concert recording from the pre-take.


Example 2: Recording from the Drøbak performance

Part 3 - Oslo

For the third and last recording we chose the Rikshospitalet in Oslo (Oslo University Hospital). What we wanted to explore more closely this time was the relationship between a music performance and the world surrounding it, and how this might register in a recording. This final instalment took place in the hallway of the hospital without the occasion of a formal concert. The idea was to remove ourselves from the isolation of an artificially quiet recording space and embrace the room that we inhabited, with its people and sounds of daily life.[2] The piano on location was a beautiful old Steinway used solely by visitors and patients. It hadn’t been kept in very good condition and we did not make any attempts to refine it (by tuning etc.), in keeping with our ideas of integrating the sound of the space. We wanted a recording of the Goldberg Variations where the sound of the room, the people, the atmosphere and the instrument would all be allowed to make an impression on the music and musician. Our goal was to make it impossible to imagine that this recording could have happened anywhere else. We wanted to go as far as possible in creating a symbiotic relationship between the music and its circumstances. The music was to influence its surroundings as well as the surroundings were to influence the music, to such an extent that even the smallest sounding detail would be an integral part of the result. Johann focused on positioning the piano within the space to best capture the instrument in a natural focus, instead using the microphones to enhance the sounds of the hospital artificially. He also chose not to include artificial reverb (which is a very common practice in classical recordings), to convey a more realistic high-resolution sound image of the music in the space.

The recording in Oslo ended up being a unique experience for both of us. I am still impressed by Johann’s beautiful capturing of the hospital and all its noises, tightly woven in with the sound of the piano. During our visits to the hospital we met several patients and spent some time interacting with them, answering questions about the project. During the performance, they came and went, remaining present in my mind while I worked my way through the piece. The decision not to isolate ourselves and to meet and interact with the people inhabiting the space was vital to the project.

In the end the third recording presents a shared timeline between the music and a location, with all its native flavours still intact. We found this refreshing, given the extended focus on purity and isolation one often encounters in classical music recordings.

When listening back to the three recordings I am inspired by the many doors they open for me as a musician. Adapting a more reflective approach to my performance practice is very intriguing. The experience of the performer as part of the interpretation itself is also something that could be a great starting point for future projects. The three recordings have given us some ideas for interesting aspects to explore. Our perception at this point is that a more conceptual approach could prove both valuable and enriching. An unavoidable next step would be to investigate the dialogue between the classical recording practice and the more experimental audio culture.

Finishing this project feels very much like the start of something new and exciting. Johann and I are eagerly on the lookout for new collaborators, venues and concepts.


Example 3: Recording from the Oslo performance



[1] In Schiff’s previously mentioned recording of the Diabelli Variations he builds two contrasting interpretations on two pianos he knows very well. While he does wonderful work drawing on the different characteristics of the instruments, he does so out of choice, staying very much in control.

[2] A project of interest in this context is Aisha Orazbayeva’s 2013 recording of Sciarrino’s 6 Caprices for Violin on the album Outside on Nonclassical. It features an interesting mix of studio and location recording, often combined within each track.