Firewire vs. USB
A great deal has been said among musicians, technicians, and online about whether Firewire is better than USB, or about which of the two standards is superior. This section is not intended to change anyone's preferences or to question what a musician or technician may experience with their hardware and find useful for their art or work. Rather, the goal is to provide some clarity when comparing two audio interfaces.
First, it is important to distinguish between differences in Firewire and USB as communication standards between devices and differences between two specific devices, regardless of which communication standard they use.
If we focus specifically on the characteristics relevant to real-time digital audio transfer, both Firewire and USB provide isochronous transfer mechanisms. These mechanisms are designed to reserve bandwidth and perform transfers within defined time intervals, making them particularly suitable for applications such as real-time audio and video. Anyway, this does not mean that they work internally in the same way. Beyond which bus an interfaces uses, many other things have a significant impact on actual performance: internal controllers, device firmware, driver, the way that the operating system handles transfers, etc.

Differences between the standards
USB was designed as a much more general-purpose bus and is used for a very wide variety of devices. In addition to real-time audio transfer, it can carry data for storage devices, input devices such as keyboards and MIDI controllers, cameras, network interfaces, and many other types of peripherals.
Firewire was also designed as a general-purpose bus and supports different types of devices and transfers. However, during the years when both standards were widely used, Firewire became particularly prominent in certain markets, including audio interfaces and video cameras.
One historically relevant characteristic of Firewire was its direct memory access architecture and its ability to perform isochronous transfers with relatively low CPU overhead. The Firewire controller could use DMA to move data between the bus and system memory, reducing the amount of work required from the CPU. But USB also uses DMA in modern host controllers.
Historically, these characteristics helped Firewire interfaces achieve small buffers and low latency with relatively low CPU overhead. However, improvements in USB hardware, host controllers, drivers, and operating systems have made it possible for USB audio interfaces to operate at very low latencies as well.
For this reason, it is no longer accurate to state that Firewire inherently provides lower latency than USB.
Bandwidth
Firewire 400 provides a signaling rate of 400 Mbit/s, while USB 1.1 Full-Speed provides 12 Mbit/s. In this respect, Firewire 400 has considerably more transfer capacity than USB Full-Speed. However, USB 2.0 High-Speed reaches 480 Mbit/s, while the first generation of USB 3.0 reaches 5 Gbit/s.
USB 2.0 (working as High-speed) has sufficient bandwidth for a wide range of applications. The Behringer X32 digital console, for instance, is a clear example of these capabilities. It can function as a high-speed USB audio interface with 32 input channels and 32 output channels, at a 48 kHz sampling rate.
Is Firewire better than USB for audio?
In general, the connection standard alone is not enough to determine whether one audio interface is better than another.
A Firewire interface can have an excellent implementation and provide very low latency, while another Firewire interface may have poor drivers or a less efficient implementation. The same applies to USB.
When comparing two specific audio interfaces, it is therefore more useful to examine their complete implementation: A/D and D/A converters, clocking, internal architecture, drivers, minimum buffer size, stability, channel count, supported sample rates, and other features.
This leads to another discussion, which is outside the scope of this section: many older audio interfaces used Firewire simply because it was the standard available at the time. Some of them may offer different features from modern USB interfaces. Determining which one is better requires comparing their characteristics as audio devices — including their conversion, preamplification, frequency response, noise, distortion, and overall capabilities — rather than simply comparing the connectivity standard used to connect them to the computer.
A note on USB versions
Since this section discusses some differences between USB and Firewire, it is also worth clarifying the relevance of the different USB generations to audio.
For most professional audio applications, USB 2.0 High-Speed already provides enough bandwidth for a considerable number of channels. USB Audio Class 2.0 allows High-Speed audio transfers with 0.125 ms intervals, while also supporting significantly higher channel counts and sample rates than UAC 1.0.
As a reference, a USB 2.0 High-Speed implementation can transfer at least 60 input and 60 output channels at 96 kHz/24-bit. The actual number of channels also depends on how samples are packed, which endpoints are used, and the rest of the implementation.
For most professional recording and playback systems, these capabilities are more than sufficient. USB 3.x primarily provides substantially greater transfer capacity, which is useful for devices that need to move large amounts of data. In audio applications, this can allow higher channel counts or certain combinations of sample rate and bit depth, but USB 3.0 does not inherently improve audio quality or automatically reduce latency.
In other words, for an audio interface with, for example, 8, 16, 32, or even many more inputs and outputs, USB 2.0 can provide sufficient capacity and pretty low latency without requiring a later USB generation.