Combined audio/MIDI recording can place high demands on a system, requiring a fast CPU and hard drive, an appropriate choice of audio driver, and a careful setting of the audio buffer size. These components affect your ability to record, process, and play sound, particularly in real time.
In Chapter 5, we introduced the subject of latency in digital audio systems. The problem of latency is compounded when MIDI data is added to the mix. A common frustration in MIDI recording sessions is that there can be an audible difference between the moment when you press a key on a MIDI controller keyboard and the moment when you hear the sound coming out of the headphones or monitors. In this case, the latency is the result of your buffer size. The MIDI signal generated by the key press must be transformed into digital audio by a synthesizer or sampler, and the digital data is then placed in the output buffer. This sound is not heard until the buffer is filled up. When the buffer is full, it undergoes ADC and is set sent to the headphones or monitors. Playback latency results when the buffer is too large. As discussed in Chapter 5, you can reduce the playback latency by using a low latency audio driver like ASIO or reducing the buffer size if this option is available in your driver. However, if you make the buffer size too low, you’ll have breaks in the sound when the CPU cannot keep up with the number of times it has to empty the buffer.
Another potential bottleneck in digital audio playback is the hard drive. Fast hard drives are a must when working with digital audio, and it is also important to use a dedicated hard drive for your audio files. If you’re storing your audio files on the same hard drive as your operating system, you’ll need a larger playback buffer to accommodate all the times the hard drive is busy delivering system data instead of your audio. If you get a second hard drive and use it only for audio files, you can usually get away with a much smaller playback buffer, thereby reducing the playback latency.
When you use software instruments, there are other system resources besides the hard drive that also become a factor to consider. Software samplers require a lot of system RAM because all the audio samples have to be loaded completely in RAM in order for them to be instantly accessible. On the other hand, software synthesizers that generate the sound dynamically can be particularly hard on the CPU. The CPU has to mathematically create the audio stream in real time, which is a more computationally intense process than simply playing an audio stream that already exists. Synthesizers with multiple oscillators can be particularly problematic. Some programs let you offload individual audio or instrument tracks to another CPU. This could be a networked computer running a processing node program or some sort of dedicated processing hardware connected to the host computer. If you’re having problems with playback dropouts due to CPU overload and you can’t add more CPU power, another strategy is to render the instrument audio signal to an audio file that is played back instead of generated live (often called “freezing” a track). However, this effectively disables the MIDI signal and the software instrument so if you need to make any changes to the rendered track, you need to go back to the MIDI data and re-render the audio. Check pdf2word.org/ for more tips.