Reply to thread

Message: [QUOTE="Bombastus, post: 91756, member: 2178"] When you play a software instrument, lowering the audio buffer reduces output latency, while larger buffers trade responsiveness for greater processing stability. A slow-feeling piano setup therefore does not automatically mean the MIDI keyboard is sending notes late. The path runs from your finger to a MIDI event, through the piano software and audio driver, then into your headphones or speakers. Key action still affects how playing feels. But [B][URL='https://goldmidi.com/community/threads/learning-piano-on-a-midi-keyboard-has-real-limits.77170/']the physical limits of learning piano on a MIDI keyboard[/URL][/B] are separate from a sound path that responds late. People often call the whole problem MIDI keyboard latency because the keyboard is where the action starts. In practice, the useful question is which stage adds enough delay for your timing to feel detached from the note. [HEADING=2]The delay often starts after MIDI has arrived[/HEADING] Start with the sound source, not the USB cable. Load a lightweight piano or basic synth, remove heavy effects from the instrument and master bus, and use wired headphones. Play the same short passage that exposed the delay. If the response suddenly tightens, your controller was not the main problem. Look-ahead dynamics, linear-phase processing, convolution effects, oversampling, and large sample instruments can all make a live software instrument less immediate. Host behavior matters too, because delay compensation can change what you hear while monitoring. Audio buffer size is the next useful control. Smaller buffers make the software hand audio to the device more often, reducing waiting time. They also give the computer less room to finish each processing block before playback. Do not read a setting such as 64 or 128 samples as your complete finger-to-ear latency. At 48 kHz, 128 samples represent about 2.67 milliseconds of audio. The full path can still include driver buffering, software processing, converter delay, and processing inside the playback device. Sample rate changes how much time a fixed number of samples represents, but raising it is not a free latency cure. Higher rates also increase processing demand. The practical target is the lowest buffer your machine can sustain without audible glitches. Windows users should also check which audio driver the music application is actually using. ASIO is common for low-latency music work, while modern Windows audio APIs can also provide low-latency paths with compatible software and drivers. Buying an interface before checking the active driver is premature. [HEADING=2]Output latency matters more than the impressive number[/HEADING] Piano practice with a virtual instrument is mainly an output-latency problem because no microphone signal needs to return through the computer before the piano sounds. Some DAWs display roundtrip and output latency separately. The output figure is the more relevant one when a MIDI controller triggers a software piano. A 2016 [B][URL='https://andrewmcpherson.org/instrumentslab-data/data/andrew/jack_am2016.pdf']action-to-sound latency study[/URL][/B] tested zero, 10, and 20 millisecond conditions on a digital percussion instrument. The 20 millisecond condition received significantly poorer quality ratings and greater difficulty than zero or 10 milliseconds. Variable delay mattered too, with 10 milliseconds plus or minus 3 milliseconds rated similarly to the 20 millisecond condition. Those findings do not give every pianist a universal cutoff. They do show why a setup can feel inconsistent even when an average latency figure looks respectable. Fast scales and repeated notes make unstable timing hard to ignore. Bluetooth can create a similar diagnostic trap. A wired MIDI keyboard paired with Bluetooth headphones may deliver note messages promptly while the audio arrives late. Changing MIDI cables or USB ports cannot repair the slow part of that path. [HEADING=2]A clean test isolates the slow stage[/HEADING] Use one plain instrument, one wired audio output, and no unnecessary effects. Set the buffer lower in small steps until the response feels right or the system produces clicks, pops, or dropouts. Then move back to the lowest stable setting. Next, compare the same passage with your normal piano library. If the basic instrument feels tight but the piano does not, inspect the library's quality settings and microphone options. Convolution processing, oversampling, or effects loaded automatically may be costing you responsiveness. A delayed attack can also survive an apparently excellent DAW latency reading. Digital speakers, wireless receivers, room-correction processors, and some monitor systems can add processing after audio has left the interface. Lowering the computer buffer will not remove delay introduced farther downstream. Keep the test boring. Change one component, replay the same phrase, and listen for whether the attack moves closer to your finger. Changing five settings at once only hides the culprit. If lowering the buffer makes no audible difference, stop chasing the buffer. Swap Bluetooth audio for wired headphones, bypass downstream processing, try a simpler instrument, and check the active audio driver. Only then does the controller itself deserve serious suspicion. [/QUOTE]

Name