Towards A Computerized Dapu Process: A History

Dapu is the analytical reinterpretation of a written passage of guqin tablature manuscript, translating it from a written piece of music. According to the Zhongguo Yinyue Cidian (Chinese Musical Dictionary), dapu is “a term referring to the process of playing out a qin melody from a qin manuscript. Since qin manuscripts do not directly record musical notes but only string positions and fingerings, as well having a considerable flexibility in rhythm interpretation, dapu’ers must be familiar with conventions in and performance technique to deduct the melodies’ progression, and then to recreate it. Its aim is to reproduce the state in which it was originally performed – since most surviving pieces are lost in practice, one must undergo the dapu process to restore the music.” (p.64)

The traditional guqin manuscript system was conceived by Cao Rou presumably around the 10th century CE. It supplemented the musician as a reminder for fingering technique in case of forgetfulness, assuming that the melody and rhythm are already impressioned into the player’s mind. As primarily an oral tradition, students studying under a master would not have to encounter the dapu process as melodies would have been passed from performance imitation, without drawing from a written medium. John Thompson posits that the tradition of learning guqin pieces directly from a book, without aid of a teacher or another musician transmitting a predetermined rhythm was nonetheless existent and popular.

Much of the discussion on dapu in guqin circles and academia discuss on whether the act of dapu is an “archaeological” rediscovery of an ‘authentic original’ performance through surmising or deducing missing elements required in musical performances from the imperfect information on existing tablature, versus a form of original artistic creation by the contemporary guqin artist derived from the historical text. Regardless of where artists may lie on this spectrum, the process of dapu remains the same.

Technically, sight-reading an original text for musical performance is possible and crudely fits into the definition of dapu as translation from text to performance, and its credibility and authenticity may be verified through various kinds of audio/visual recording devices. However, as a scholarly tradition, the legitimacy of the translation is recognized only through the quantifiable, empirical medium of the text – in other words, while the common definition iterates dapu’s ultimate purpose is for the music to ‘transcend’ or ‘liberate’ from the textual realm into the aural one, in reality it never escapes this plane of existence but rather becomes increasingly confined by the empirical scrutiny of increasingly meticulous annotation and establishment of parameters.

Hence, dapu in the modern sense means the meticulous recording or revising of necessary components to guqin music, which involves the use of jianzipu tablature, as well as another system that records pitch and rhythm. The common method involves three major steps:

1. rewriting the tablature in compatible layout with the score for the pitch and rhythm (typically from vertical layout to horizontal layout, when dealing with historical sources),
2. determine the number of sounds and correlate the characters to each of its pitches according to tuning, accounting for any mode changes should a solfege-based recording system (e.g. jianpu) is used, and
3. assign and record the tempo and rhythm of the performance, based on the dapu’er’s deliberation.

At any point during these three steps, the dapu’er also has to make decisions and be held accountable for discrepancies and changes between the source text and the new score, whether they’d be intentional or accidental. For example, one may discover that a cadenza may not end in a typical matching paired interval but one pentatonic tone apart, that can be easily corrected by using the subsequent string. The artist will then have to deliberate on whether that is intentionally required by the piece, or if it is a misprint due to error in its transmission (such as a scribal mistake or faulty photocopying). If such a deliberation is made, regardless of whether an actual change is made from the original text – it is also up to the dapu’er to record such changes in writing to be held accountable for future reference and scrutiny.

A flowchart of how the scores are produced in Standards of the Guqin. The final product is a PNG image.

Even in the early 21st century, dapu for the most part is a manual process – scores are written and rewritten by hand, and the pitches linked up by hours of tedious cross-referencing with the instrument’s pitch charts. Although publisher-grade scores now involve computer-assisted drawing, the musical work itself remains unchanged. Taking my work Standards of the Guqin as example, the tablature is rewritten digitally by a database of prearranged vector images of strokes that resemble parts of jianzipu that are formed together by entries that group them together, as well as separately creating the five-line staff music using a market-available typing and output program for such purpose.

To date, there is no available software (commercial or open-sourced) that can perform this task because the technology to digitize jianzipu itself is young and tentative. With the exception of Guangling Shenqi developed by guanglingsan.com, all other projects struggle along how to account for all possible variations, much less categorize and designate a dynamic input system. Once that logistical issue is resolved, the next logical step is for the computer software to parse the text as a generated sound and pitch. To effectively actualize this, the computer-generated jianzipu previously mentioned will need profiles of each character with a breakdown of the necessary parameters for the computer to process.

A typical parsing of an actual notation character will follow this heuristic: First, determine the current tuning of the piece. Based on the pitches of the open strings, all given pressed or harmonic positions are extrapolated from there, and it reduces the chance of miscalculation or inflexibility to external tuning. Next, determine if the left hand information is an open, pressed, or harmonic note. Once these pieces of information are gathered, a pitch (and type of tone) can be derived, but it would be up to the musician to mold the notes into a sensible piece of music by the addition of rhythm, and optionally annotations on performance style. When the missing elements are edited into the new score, a computerized printout or rendition will forego the need of another manual rewrite for spacing and cleanliness as notes are automatically spaced and formatted to the page.

Currently, some experimental projects in China also explore in developing a ‘fuzzy AI’ in allowing the computer to determine the missing rhythms and other stylistic treatments – in other words, to fully automate the process. This objective is no smaller than developing an independent thinking computer that can freely compose a musical masterpiece, and belongs to a different category from the scope of developing a digital guqin manuscript archive and dapu assistant mechanism. Another pursuit is in Optical Character Recognition (OCR) technology integration with guqin tablature to allow speedy archive and digitization of the classic texts, which may speed up the process greatly as compared to data entry using existing techniques (ie. typing characters in long form).

Commercially, digitally simulated guqins are marketed mostly as novelty apps or addons to larger virtual realities. While these show promise for a market-ready dapu software, the final step has yet to be taken, and measures have yet to be perfected.

Further reading: Zhou Changyue. Guqin Yishu de jiqi yanyi (Mechanical interpretation of the art of guqin). Beijing: China Science Publishing. 2013. Table of Contents available at http://www.baiyue-music.com/song_con.php?idept=2&isdept=7&pk=2470&page=

Chang Hong. “MIDI Guqin yanzou: Kexue yu yishu de chuangxin jiehe (MIDI Guqin performance: A culmination of science and artistic innovation),” People.com.cn, snapshot backup at http://swannbb.blogspot.ca/2015/01/digital-guqin-at-beginning-chen.html. Last accessed September 17, 2015.

Zhang Yingxue. “ChinAR,” http://www.yingxuetzt.com/#!project01/c1mqf. Last accessed September 17, 2015.