What is NotePerformer?
NotePerformer by Wallander Instruments [1], which I have used occasionally for some time, discontinued its support for external VST3 playback engines as of version 5.1.2 (according to the official announcement).
From my perspective, this feels like a natural direction. However, for those who relied on external libraries, it may be a somewhat disappointing change.
Going forward, I hope that NotePerformer will further refine its own score-interpretation engine.
At present, alongside my work on Beethoven’s Piano Sonata No. 29, I am also programming the third movement of String Quartet No. 15. When I load the score into Dorico and play it back with NotePerformer, I am once again reminded how useful it is for grasping the overall musical flow and structure.
In the past, I regarded NotePerformer as “a playback engine that sounds reasonably good.” Today, however, it is more appropriate to understand it as an engine that reads the score.
In this article, I would like to organize my own understanding of how NotePerformer works conceptually. Although this discussion focuses on strings, NotePerformer is actually a full orchestral playback engine that covers a wide range of instruments.
I looked for detailed technical documentation, but beyond the official manual, very little has been published. Therefore, this article is based on developer comments found in various forums as well as observations of actual behavior, with some degree of informed speculation.
[1]: NotePerformer
What Does NotePerformer Actually Do?
The first key point in understanding NotePerformer is that it does not simply play back notes in sequence.
Conventional sound libraries—especially string libraries used in DAWs—generally reproduce MIDI data as it is. Dynamics, timing, and phrasing nuances must all be provided explicitly by the user through control data such as MIDI CC.
In contrast, NotePerformer analyzes the score itself and automatically generates musical expression from it.
For example, it interprets information such as:
- Presence or absence of slurs
- Crescendos and diminuendos
- Relationships between voices
- Phrase structure
and appears to internally determine how the music should be performed.
In other words, NotePerformer is best understood not as “a sound library that plays notes,” but as “an engine that reads the score.”
The Concept of Lookahead
At the core of this system lies what is known as lookahead.
This mechanism has been present since early versions, but its accuracy and behavior have improved over time. In particular, from Version 3 onward, playback became more natural, especially in integration with notation software such as Dorico and Finale.
Traditionally, lookahead has been used mainly to compensate for latency in sound production. In string libraries in particular, legato transitions tend to introduce delays, making it necessary to process upcoming notes slightly in advance in order to maintain proper timing.
In fact, in physically modeled and hybrid instruments such as the SWAM engine by Audio Modeling and Sample Modeling instruments, similar mechanisms—effectively a form of lookahead—have long been implemented internally to ensure consistency between real-time performance and articulation.
Later, Tokyo Scoring Strings introduced a “Lookahead mode” for sampled instruments, aimed at reducing the need for manual timing adjustments in DAWs caused by legato delays (often referred to as transition speed or slur rate). This development played a significant role in bringing such functionality into wider use.
Today, however, the role of lookahead has evolved considerably.
In the case of NotePerformer, it likely references several measures ahead to determine:
- where a phrase is heading
- Where to Breathe.
- Which voice part is brought forward?
This is no longer merely a matter of timing correction.
This is not mere timing compensation, but rather "a process that determines the flow of the music itself.
A Hybrid Sound Engine
Another key feature of NotePerformer is that it uses a hybrid sound engine.
Officially, it is described as combining a small set of high-quality samples with modeling techniques.
Typical string libraries rely on large numbers of samples for different dynamics and articulations, often requiring tens of gigabytes of storage.
By contrast, NotePerformer uses only a minimal set of samples, while generating elements such as
- Volume change
- vibrato
- Sound connection
in real time.
As a result, it achieves a certain level of musical expressiveness with a footprint of less than 1 GB.
How It Differs from Other String Libraries
To better understand its position, it is useful to compare NotePerformer with other string libraries.
Most string libraries can be broadly divided into two categories.
One is sample-based libraries such as LASS, MSS, and TSS. These rely on MIDI input, with the user shaping timing, dynamics, and phrasing manually.
The other is modeling-based instruments such as SCES and SWAM, which aim for more continuous expression through real-time control of performance parameters.
In both cases, the focus is on “how to perform.”
By contrast, NotePerformer focuses on “what is being expressed in the score,” constructing musical output directly from notation.
Writing the Score Becomes Sound Design
One of the most striking aspects of using NotePerformer is how strongly the playback depends on the quality of the notation.
If the score is
- ambiguous in slurring
- lacking dynamic markings
- poorly organized in terms of voice leading
the playback will reflect that ambiguity.
Conversely, when
- The phrasing is clear and
- The roles of the voice parts are organized
the result can be surprisingly musical.
n other words, “The precision of the score directly determines the quality of the sound.”
The Role of NotePerformer
From all of the above, it becomes clear that NotePerformer occupies a somewhat different role compared to conventional sound sources.
- DAW Sound Source: Creating a Performance
- Modeling Sound Source: Creating Sound
- NotePerformer: Reconstruct music from sheet music
This framework helps to clarify its position.
Perhaps licensing issues also played a role, but with the discontinuation of VST support, NotePerformer seems to be moving toward a purer form as a “score-interpretation engine.”
This may represent an alternative approach to music production—distinct from the traditional process of “creating sound.”
It is also worth noting that the idea of generating performance from a musical score is not unique to NotePerformer.
I have previously come across a project called “Melisma,” [2] which generates performance from MusicXML. It appears to share a similar direction in that it attempts to construct music directly from notational information.
I will return to this topic later in more detail.
[2]: Melisma
In the next article, I will take a closer look at lookahead, one of the key elements behind this system.
Japanese version is abailable here.