Digital Audio Computing Infrastructure

数字音频被广泛用于从生物学领域研究到视频游戏的研究领域，然而支持这些研究的软件平台在很大程度上是不兼容的，通常设计糟糕，而且很少可移植。软件缺陷和未实现的功能阻碍了新用户，研究人员浪费了大量的资源来重复工作。在这个项目中，PI将创建一套协调和可互操作的软件库和应用程序，使研究人员能够轻松访问音频数据。低级音频接口将满足对设计良好、跨平台、开源库的重要需求，这些库提供对音频数据的直接访问。将开发用于音频输入和输出、访问音频声音文件、MIDI输入和输出以及访问标准MIDI文件的库。这项工作不会创建新的硬件或设备驱动程序，相反，它将在现有操作系统音频api之上添加一个薄抽象层，从而隐藏难看的细节和操作系统依赖性，使音频访问更简单。高级应用程序将支持需要通用音频可视化工具的用户，该工具提供了当前不存在的音频分析和注释工具。音频可视化功能将包括波形、频谱图、其他1D和2D时间功能、文本和图形注释，以及音乐数据的钢琴滚动显示。注释功能将允许用户在显示器上绘制草图，编辑图形叠加，并以简单的基于文本的表示形式保存注释数据，以便与其他软件交换。分析软件将集成广泛用于音频分类研究的Marsyas项目；事实上，PI将尽可能地建立在现有软件的基础上，既减少了总工作量，又通过利用已建立的社区和忠诚度来增加影响。更广泛的影响：这项工作将建立事实上的标准，并使不同的研究领域受益，不仅包括音乐和音频技术，还包括人机界面、辅助技术、声学、人类感知、监视和安全、患者监测和护理以及教育等。项目成果将促进应用程序中音频的整合，并通过允许跨平台共享音频软件来减少重复工作，从而使教师更容易将“动手”音频处理纳入其课程，并支持与普遍访问和协作系统相关的创新努力。

Digital audio is widely used for research in areas from biology field studies to video games, yet software platforms to support such research are largely incompatible, often poorly designed, and rarely portable. Software bugs and unimplemented features discourage new users, and researchers waste vast resources duplicating efforts. In this project the PI will create a suite of coordinated and interoperable software libraries and applications, giving researchers easy access to audio data. Low-level audio interfaces will fill an important need for well-designed, cross-platform, open-source libraries that provide direct access to audio data. Libraries will be developed for audio input and output, access to audio sound files, MIDI input and output, and access to standard MIDI files. This effort will not create new hardware or device drivers, rather it will add a thin abstraction layer above existing operating system audio APIs that hides the ugly details and operating system dependencies, making audio access simpler. A high-level application will support users who need a general-purpose tool for audio visualization that provides instruments for audio analysis and annotation that do not currently exist. Audio visualization capabilities will include waveforms, spectrograms, other 1D and 2D functions of time, text and graphical annotations, and piano-roll displays of music data. The annotation facility will allow users to sketch on displays, edit graphical overlays, and save annotation data in simple text-based representations for exchange with other software. Analysis software will include an integration of the Marsyas project, which is widely used in audio classification research; indeed, the PI will build upon existing software wherever possible, both to reduce the total effort and to increase the impact by tapping into established communities and loyalties. Broader Impacts: This work will establish de-facto standards and benefit diverse research areas including not only music and audio technology but also human-computer interfaces, assistive technologies, acoustics, human perception, surveillance and security, and patient monitoring and care, and education, to name but a few. Project outcomes will both facilitate the incorporation of audio in applications, and reduce duplicated effort by allowing audio software to be shared across platforms, thereby making it easier, for example, for teachers to incorporate "hands-on" audio processing into their curricula, and supporting innovative efforts relating to universal access and collaborative systems.