Mobile DLS规范1.0a：下载声音格式详解

需积分: 5 42 浏览量更新于2024-06-16 收藏 568KB PDF 举报

"Mobile DLS (Downloadable Sounds Format for Mobile Applications) 是由 MIDI Manufacturers Association (MMA) 发布的一种音频格式规范，版本1.0a，主要基于DLS 2.1规范，用于移动设备。该文档详细描述了Mobile DLS的特性，并指出与DLS 2.1的不同之处，以便读者可以直接理解Mobile DLS，而无需参考DLS 2.1的完整规格。在1.0a版本中，进行了两个主要修订：删除了结构体_POOLCUE中的ULONG ulEntryIndex字段，并修正了FOURCC_wavemmio定义的字母大小写问题。" Mobile DLS是MIDI的一个扩展，旨在为移动应用提供下载式声音格式。这个标准允许用户在移动设备上存储和播放高质量的音频内容，特别适合于有限存储空间的环境。Mobile DLS是从DLS（Downloadable Sounds）2.1版本发展而来，后者是一种广泛应用于桌面系统的MIDI声音扩展技术。 DLS 2.1规范允许音乐应用程序通过网络下载乐器样本和其他声音效果，然后在本地系统上使用。Mobile DLS则针对移动设备进行了优化，可能包括内存管理、文件大小压缩以及对电池寿命的影响等方面的考虑。由于Mobile DLS与DLS 2.1高度相似，但存在差异，因此文档中会明确指出这些差异，以帮助开发者准确实现和使用Mobile DLS功能。在1.0a版本的修订中，文档去掉了`POOLCUE`结构体内的`ULONG ulEntryIndex`字段。这可能是为了简化结构体，减少不必要的数据存储，从而优化移动设备的内存使用。同时，修正了`FOURCC_wavemmio`宏定义的字母大小写问题，确保代码与标准约定一致，提高代码的可读性和兼容性。 Mobile DLS规范的发布，对于移动游戏、音乐制作应用和其他需要高质量音频内容的移动应用程序来说，是一个重要的进步。它为开发者提供了标准化的框架，使得音频内容可以在各种不同的移动平台上以一致的方式运行和播放。然而，由于MMA声明他们不对文档或规范的使用产生的任何后果负责，所以开发者在采用Mobile DLS时需要自行评估风险并确保符合所有相关法律法规。 Mobile DLS是MIDI生态系统中针对移动设备优化的声音格式，它建立在DLS 2.1的基础上，提供了一套适应移动环境的音频处理和播放标准。通过了解和遵循Mobile DLS规范，开发者可以创建更高效、音质更佳的移动音频应用。

DOWNLOADABLE SOUNDS FOR MOBILE APPLICATIONS

Version 1.0a Page 9 November 2004

1.5 Digital Audio Engine

The digital audio engine consists of a Digital Oscillator, Digitally Controlled Filter, and Digitally Controlled

Amplifier.

1.5.1 Digital Oscillator

The Digital Oscillator takes as its input an audio sample data and plays it back at a rate controlled by the articulation

data. It must perform interpolation on the sample data in order to convert it from the sample rate at which it was

recorded to the output sample rate, applying any pitch translation required by the articulation data.

The sample defines whether it is to be played once through or looped. If looped, the sample supports one looped

region, as defined by a pair of fixed sample data point indices. The loop start specifies the first sample data point in

the loop, while the loop length specifies the number of samples in the loop. There is no fractional component to the

loop points.

Mobile DLS supports the Forward Loop type shown in Figure 3 and the Loop and Release type shown in Figure 4.

In the Forward Loop, the device begins playback at the start of the sample. Upon reaching the end of the loop as

defined by the loop length, the device returns to the start of the loop. Playback continues in this fashion until the end

of the release segment of the envelope.

In the Loop and Release loop, the device begins playback as in the simple loop form. However, at the time that the

MIDI Note-off event is received, the device does not return to the beginning of the loop, instead playing straight to

the end of the sample. It is very important that the device not jump directly to the release segment, but continue

playing from wherever it may be in the loop straight through to the end of the sample, thus maintaining continuity in

the samples.

Transient Loop

Loop Start

Loop Length

Figure 3 — Forward Loop

Transient Loop

Loop Start

Loop Length

Release

Figure 4 — Loop And Release

The sample structure defines the MIDI note number, fine tune, and sample rate of the recorded sample, providing all

the information required for the digital oscillator to pitch shift the sample appropriately for the intended pitch. The

Digital Oscillator must be capable of pitch shifting the sample with a range of up two octaves and down four octaves.

This pitch shift is the sum total of all tuning parameters, including sample fine tune and MIDI pitch bend.

DOWNLOADABLE SOUNDS FOR MOBILE APPLICATIONS

Version 1.0a Page 10 November 2004

Description Mobile DLS instrument data

Loop type Forward Loop and Loop and Release

Waveform data format RIFF WAVE PCM format

Table 4 — Supported Digital Oscillator Functionality

1.5.1.1 PCM Sample Data Format

When the wFormatTag is equal to WAVE_FORMAT_PCM, the sample data can be 16-bit two's complement or 8-bit

offset two's complement (often referred to as 8-bit unsigned data, with zero represented as 0x80). This complies with

the both the RIFF WAVE PCM format and common practice on the Macintosh. In order to use the 8-bit format, it is

usually necessary to convert it to standard two's complement. To convert between offset two's and standard two's

complement (signed 8 bit, -128 to +127), XOR each byte of the source sample value with 0x80.

1.5.2 Digitally Controlled Filter

The input to the Digitally Controlled Filter is the audio data stream produced by the Digital Oscillator. The input

samples contained in the audio data stream are then filtered as required by the articulation data. The filter is defined

to be a 12 dB/octave cutoff slope lowpass digital filter with a transfer function of the form:

This transfer function represents a single complex pole pair located within the unit circle at radius r and angle θ

where:

= -2 r cosθ

= r

K = g (1 + b

+ b

)

where g is a gain factor which depends on the intended resonance. This is done in order to keep the perceived

loudness approximately constant. Otherwise, the addition of resonance will make the sound louder. As the

resonance is increased, g is decreased as follows:

where the resonance is expressed in decibels. Another way to look at this is that the DC gain of the filter is attenuated

by a factor equal to half the decibels of the resonance value.

When g is expressed in decibels, the relationship between g and resonance is:

For example, if the specified resonance is 20 dB, g

is –10 dB. Since g is multiplicative to the numerator of the

transfer function, it is equivalent to apply it after the filter by summing g with the other gain sources for the signal.

However, it is desirable to perform this within the filter when bit-width becomes a limiting factor to headroom within

the filter delay memory.

Such filters can either express a non-resonant low-pass filter with a frequency response similar to Figure 5, or a filter

with a marked resonant peak with a response similar to Figure 6.

)(

−−

zbzb

resonance

−

resonance

−

DOWNLOADABLE SOUNDS FOR MOBILE APPLICATIONS

Version 1.0a Page 13 November 2004

So, θ can thus be expressed as a function of F

and r:

The resonance of the filter is given by the following expression:

where r and θ are the radius and angle of the poles of the filter. Note that this is decibels, not absolute gain units.

Given this relationship, the radius can be calculated as a function of resonance and θ as shown in the following

equation:

where resonance is specified in decibels. The derivation of these equations is left as an algebraic exercise for the

reader.

1.5.3 Digitally Controlled Amplifier

The final module in the signal path is the digitally controlled amplifier (DCA). It modulates the amplitude of the

digital oscillator signal, responding to controls from the articulation section, of which the amplitude envelope is

probably the most important.

For the purposes of this document, the maximum output of the DCA is assumed to be 0 dB or unity gain, though as a

practical matter, some attenuation may be required prior to summing all the oscillator outputs together to form the

final synthesizer output. Therefore, all changes in the DCA are measured in dB units of gain from the 0 dB reference

point, and all units of dB in this document refer to voltage, rather than power. Positive units of gain represent an

increase in output voltage, and negative units represent a decrease in output voltage. The MIDI pan signal is fed

through a transform that converts to left and right amplitude signals. These signals are fed separately into the DCA to

generate stereo output from the monophonic signal entering the DCA.

The gain of the DCA

can be varied dynamically through the use of modulators. Since the steps in the input sources

can be rather severe, the designer must minimize audible artifacts (such as "zipper" noise) by filtering the control

source. This can be accomplished by interpolating between control events (such as MIDI controller sources).

1.6 Articulation Modules and Connections

The Articulation modules include two low frequency oscillators (LFOs), two envelope generators, and a number of

MIDI controller inputs. A configuration is defined by connecting this base set of modules and setting the controls for

each one.

1.6.1 Signal Flow

The Control Logic module translates MIDI note and sustain events into region/sample choice and associated control

signals, including Key (the note number), Velocity (note emphasis), and Gate. Gate defines the start and end times of

(

)

(

)

(

)

( )













−

−−+++−

2444224

2562

125612161

arccos

cscs

FFr

rFrFFrF













−

+−

sin)1(

cos21

log20

resonance

sin

110sincos

20/

10/

−+

θθ

resonance

剩余92页未读，继续阅读

weixin_36989426

粉丝: 0
资源: 20

Mobile DLS规范1.0a：下载声音格式详解

改进的RNNPT算法：多角度MDLS的高精度颗粒粒度分布反演

单片机与MDLS液晶显示接口及LMA97S005AD模块设计

多角度MDLS中的基线误差补偿提升颗粒分布反演精度

file-metadata:在macOS上使用`mdls`获取文件元数据

在macOS上使用mdls获取文件元数据

液晶显示模块MDLS16268C在校验信号源设计中的应用

文件泄漏利用工具: ds-store-exp

tesla-title:根据选项代码获取特斯拉车辆标题

linux-帮你从微信mac客户端中抓取出所有表情

多角度动态光散射粒度分布递归非负Phillips-Twomey算法

最新资源