|
← 判断力与美学 ← Judgment & Aesthetics
SAE 判断力与美学 · 余项之美
SAE Judgment & Aesthetics · Beauty of the Remainder
2026-07-23

谭吉姆:数字音乐从未真正调过的那根弦

Tānghīm: The String That Digital Music Never Tuned

Han Qin (秦汉)

Khyam Allami 是伊拉克裔英国音乐人,任职于贝鲁特美国大学音乐智能实验室。2026 年 7 月 21 日,他在贝鲁特发布了两件互相依存的工具:数字阿拉伯马卡姆档案库(DiArMaqAr)与调律插件 Tānghīm。前者是多语言、开源的网络平台与 REST API,收录了从 9 世纪 al-Kindī 的弦律比例到 20 世纪开罗音乐大会调律标准的数百种历史调律系统,涵盖各音程(ajnās)、曲调模式(maqāmāt)和转调路径,每条记录均附有完整的文献溯源。后者是免费、开源的 MIDI 插件(VST3/AU/CLAP,支持 macOS、Windows 与 Linux),直接从档案库 API 拉取数据,让任何 DAW 里的软合成器都能调用这些调律系统。Tānghīm 支持三种调律方法:MTS-ESP(自动同步所有兼容插件)、MPE 多复音音调弯曲,以及兼容 1983 年以来全部 MIDI 硬件的单声道音调弯曲。插件界面上有十二个可拖拽的调律滑块,每个对应色谱的一个半音步,附带卡槽标记,让演奏者可以随时边听边调。

要理解 Tānghīm 的余项性,必须先看清那个已构(已凝固的结构)是什么。1983 年 MIDI 规范将"音高"定义为 0 到 127 的整数,等间距,每个半音恰好 100 音分——这是一个将西方十二平均律(12-TET)设为宇宙默认值的数学决定。此后四十年,所有 DAW、所有软合成器、所有键盘控制器都在这个已构之上运转:音高是整数,八度是十二个等距步骤。阿拉伯马卡姆音乐理论存在了逾千年,有独立的两个"书院"音名体系(从 rāst 经 dūgāh 到 sīkāh,贯穿两个 dīwān),以及历代理论家(al-Fārābī、Ibn Sīna、Mikhā'īl Meshshāqa 等)留下的精确音程比例。但这一切从未进入数字乐器的基础设施层。1980 年代第一批数字阿拉伯键盘出现后,音乐人发展出的权宜之计叫 tasyīk(或 tashrīq):把某个音降低 50 音分,制造一个粗糙的"四分音"近似。这个权宜做法本身成了新的已构——"数字阿拉伯音乐"的默认形态,用一个固定偏差代替了完整的调律身体。凿构(chisel)在两个层面同时发生:DiArMaqAr 把千年文献中的调律数据形式化为可查询的 API,Tānghīm 将这些数据植入 DAW,以三种调律方法绕过了 12-TET 地板。

被凿出来的余项是双重的。第一个余项是阿拉伯马卡姆的完整调律身体:不是 tasyīk 的 50 音分简化,而是从 al-Kindī 到开罗音乐大会历代理论家测量、命名、传递下来的完整音程结构——始终在演奏实践中存在,但始终在数字基础设施层面不可见。Tānghīm 让这个余项第一次在任何 DAW 里可见、可用、可调。第二个余项更微妙:马卡姆不是固定音高的集合,而是活的调律实践,演奏者根据转调、情境和感受不断微调每个音的音高。这是马卡姆理论的核心——"谭吉姆"(tānghīm,التنغيم)本身指的正是这个行为:音调的发出,音高关系的设定,一种活的定调术。它与 dūzān(دوزان,演奏前对弦的机械调弦)是不同的事。所有静态调律工具——Scala 文件、MTS-ESP 预设、调律表——从来只做 dūzān,不做 tānghīm。十二个可随时拖拽的滑块隐约指向了这个缺口:调律不是锁定一次的参数,而是每次演奏都可以重新校准的弦。

从 SAE 凿构循环的角度看,现在是看见这个余项的关键时刻,也可能是最后时刻。AI 音乐生成工具几乎完全以 12-TET 西方语料库为训练数据,正在把"AI 能生成的音乐"等同于"符合十二平均律的音乐"。这个等式一旦在大众意识中凝固,阿拉伯马卡姆就不再是一个尚未被处理的余项,而是一个需要另行说明的"例外"——留待某个模型的后续更新去处理。Tānghīm 在这个时机点出现,不是为了进入 AI 音乐的叙事,而是在那个叙事封闭之前,让数字工具第一次真正处理了千年调律实践的全部结构。命名缺口此刻依然敞开:"微音阶"是以 12-TET 为基准的西方词汇;"世界音乐插件"是另一种回避;"调律软件"遗漏了活的调律行为。CDM 编辑 Peter Kirn 写道:"计算机本可以做得更好;我们本可以做得更好。"这不是分类名称,而是对一个余项的承认:它一直在那里,只是从未被调过。

tanghim.app ↗

Khyam Allami is an Iraqi-British musician and researcher at the Music Intelligence Lab, American University of Beirut. On July 21, 2026, he launched two interlocking tools in Beirut: the Digital Arabic Maqām Archive (DiArMaqAr) and the Tānghīm MIDI plug-in. DiArMaqAr is a multilingual, open-source web platform and REST API documenting the full body of Arabic maqām theory — from al-Kindī's 9th-century string ratios through the 1932 Cairo Congress tuning decisions, covering hundreds of tuning systems (tanāghīm), intervals (ajnās), melodic modes (maqāmāt), and modulation pathways, each with complete bibliographic attribution. Tānghīm (free, open-source, VST3/AU/CLAP for macOS, Windows, and Linux) queries DiArMaqAr's API and brings this data into any DAW. Three tuning methods — MTS-ESP for simultaneous automatic retuning of compatible plug-ins, MPE for polyphonic pitch-bend instruments, and Mono Pitch Bend for any MIDI hardware back to 1983 — ensure nothing is out of reach. The interface is twelve draggable per-chromatic-step sliders with snap markers, so a player can adjust a maqām tuning by ear, in real time, while playing.

To understand why this is 余项之美, you need to see the already-construct clearly. The 1983 MIDI specification encoded pitch as integers 0–127, equally spaced, 100 cents per semitone — a mathematical decision that installed Western Equal Temperament (12-TET) as the universal default for all digital music. Every DAW, synthesizer, and MIDI controller has operated on this floor for forty years: pitch is an integer, an octave is twelve equal steps. Arabic maqām theory — with over a thousand years of documentation, its own two-dīwān note-naming system spanning rāst through dūgāh to sīkāh, and precise interval ratios compiled by al-Fārābī, Ibn Sīna, and Mikhā'īl Meshshāqa — was never absent from performance practice; Arabic musicians never stopped playing maqāmāt. But it was constitutionally absent from digital infrastructure. When digital Arabic keyboards arrived in the 1980s, musicians developed a workaround called tasyīk or tashrīq: lower one note by 50 cents, approximate a "quarter tone." This shortcut became its own construct — the default form of digital Arabic music, trading full maqām intonation for a single fixed deviation from 12-TET. The chisel here operates on two levels: DiArMaqAr formalizes a thousand years of tuning scholarship into a queryable API; Tānghīm inserts that data into any DAW through three tuning methods that bypass the 12-TET floor without asking the host software to change.

The remainder excavated is double. The first is the full intonational body of Arabic maqām — not the 50-cent tasyīk approximation, but the precisely documented interval structures of al-Kindī, al-Fārābī, Mikhā'īl Meshshāqa, the Cairo Congress — always alive in practice, always absent from digital infrastructure. Tānghīm makes this visible and usable inside a DAW for the first time. The second remainder is subtler: maqām is not a fixed set of pitches but a live practice of intonation, where the player continually adjusts pitch in response to modulation, context, and musical feeling. This is what the name Tānghīm names — tānghīm (التنغيم) is the Arabic maṣdar for the act of intoning, the living quality of pitch-setting, as distinct from dūzān (دوزان), the mechanical tuning of strings before you play. Every static tuning tool before it — every Scala file, every MTS-ESP preset — has done only dūzān: set the pitches and lock them. Tānghīm's twelve draggable sliders point at what remains beyond that: the intonation that happens in the playing itself.

From the SAE chisel-construct cycle, the timing matters. AI music generation tools — Suno, Udio, and others — are almost entirely trained on 12-TET Western corpora. As this equation consolidates (what AI generates = what fits 12-TET), Arabic maqām stops being an undigested remainder and becomes a labeled exception — something to request in a future model update, something to accommodate later. Before that accommodation closes the naming gap, Tānghīm arrives as the first tool to bring the full structural body of maqām theory into the digital audio production layer — not as a sample pack, not as a microtonality experiment, but as a live archive with an API and a real-time intonation interface. The naming gap is still wide: "microtonal" treats 12-TET as the baseline; "world music plugin" is the colonialist frame; "tuning software" misses the liveness. Kirn writes: "The computer can do better; we can do better." That is not a category name. It is an acknowledgment of a remainder that has been there, below the equal-temperament floor, for over a thousand years — and that digital music has, until now, never truly tuned.

tanghim.app ↗