The MAQ Index is a groundbreaking analysis of audio quality on mobile devices. It provides developers with insights about the devices best positioned to run audio apps, and it offers consumers granular information about the audio performance of their devices. The index is based entirely on performance analysis of JUCE-based apps across the iOS and Android platforms. JUCE is ROLI’s C++ framework for high-performance audio.

Each device is analyzed rigorously according to the following parameters:

Sessions

• The minimum number of sessions is 30
• Sessions that last between 3 minutes and 3 hours are considered

Criteria

• Glitchiness: The glitchiness value is an average of underruns per second, normalized on a scale of 0 to 1
• Latency: Latency is calculated as the average of (buffer size * number of buffers) / (sample rate) for all sessions on the device
• User ratings: User ratings are grouped in three categories: High, Medium, and Low. A machine-learning algorithm considers these categories alongside statistical results of glitchiness and latency on each device in order to help define the quality of user experience

Regularly tested and updated, the Mobile Audio Quality index results are presented in a data table and a Visualizer. Both formats plot performance scores according to the primary criteria of latency and glitchiness.

Scope of the data and limitations

• The MAQ is a statistical analysis of user generated data. It is not a standalone test run on devices in laboratory conditions
• Real-world conditions — such as the receipt of emails or multiple applications running in the background — can affect audio performance on devices during individual test sessions. The data presents a composite view of audio performance on a device in real-world conditions
• Glitchiness can be affected by user behavior
• Tablets, i.e. the mobile devices with the largest screens, may have glitches independent of CPU capabilities
• Latency analysis focuses on the average latency of the application based on the parameters set within the application (buffer size, number of buffers, and sample rate). It does not include input latency (touch or MIDI inputs) or output latency (Bluetooth). The actual latency of each device will be higher than the data indicates

While the MAQ Index shares granular performance scores of over 100 devices, the minute differences between devices with similar ratings usually do not affect the quality of user experience. For this reason the Index categorizes performance according to a three star system that highlights major differences in audio quality.

*** A three-star rating indicates that latency and glitchiness are negligible or virtually nonexistent. These devices are suitable for powering professional live music performances. This is the highest rating.

** A two-star rating indicates that the device is reliable for casual live music performances. Some glitches may occur, and latency may be noticeable.

* A one-star rating indicates that the device is not optimal for live music performances. Glitchiness and latency may be pronounced. However, these devices can likely be used for recording and other music-making activities that don’t require high-quality audio performance in real time.

Visualizer

The following Visualizer plots the performance of key mobile devices based on the key criteria of normalized glitchiness and latency.

Mobile Audio Quality Index Tables

The following tables rank iOS devices and Android devices.

iOS devices (alphabetical order)

Android devices (ordered by MAQ)

Definitions Audio Latency

Audio latency is calculated as the minimum latency of the audio playback, based on the recommendations of each device manufacturer. It does not consider the hardware latency of the digital analog converter, or of any other software layer of a device that may be adding latency. The latency is calculated from the number of buffers used, the hardware buffer size and the sample rate. The actual device latency will always be higher than the value we calculate as the hardware also has an associated latency, which will vary from device to device.

Audio Glitchiness

Audio glitchiness, or crackling in the audio output, is a function of the number of underruns that occur in a certain time unit during audio processing. An underrun occurs if the time it takes to process a block of audio is longer than the time it takes for the audio hardware to play back the same block. Mobile devices use techniques such as double, triple or quadruple buffering to avoid a “glitch”. Multiple buffering increases the audio reliability but it also increases latency. If the audio processor never detects an underrun, then the performance score prior to latency calculation will be 100. If the audio processor is constantly underrunning then the performance score is zero.

Session

A unique engagement with a JUCE app on Android or iOS.

User ratings

A user rating is a qualitative comment left by users on the Play Store of Apple Store. We use User Ratings to validate the statistical data.