Skip to main content

​
Sampling from audio input

Use direct recording when you want to bring in a phone, synth, turntable chain, microphone, or any other live source.

​
When to use it

  • you want original material instead of browsing files
  • you want to capture a phrase and turn it into pads or chops immediately
  • you be digging in the crates

​
Before you start

  • choose the correct input device in Settings
  • confirm the input channel selection
  • check output playback first so you can review the result immediately

​
Basic flow

1

Open the recording flow

Press Enter on an empty channel then r for audio record in.
2

Arm and capture the source

Start recording with a clean input signal and avoid clipping.
3

Review the take

Listen back before committing it to the session. Use Space to replay, and N to toggle normalization if the level needs a boost.
4

Keep or discard

Keep only takes you would actually use. The goal is fast iteration, not archiving everything.

​
Trigger recording

Trigger mode is for capturing a live source the moment it gets loud enough to use. Set the threshold just above the room noise or device idle level, then arm the recorder. The take begins at the first audio frame that exceeds the threshold. Audio before that crossing is monitored only and is not saved into the recording. Use Trigger when you want sampler-style starts from a synth, turntable, phone, microphone, or any other source where pressing [SPACE] manually would catch too much lead-in.

​
Normalizing during review

After stopping a recording, press N during the review phase to toggle peak normalization. This scales the entire sample so the loudest peak reaches -0.1 dB (0.99 linear), which is useful when:
  • the input was recorded at a conservative level
  • you want the sample at a consistent loudness before assigning it to a pad
  • you are comparing takes and want them at roughly the same volume
Press N again to revert to the original levels. The status bar shows the original peak in dB so you can see how much gain was applied. If the recording is near-silent, normalization is skipped and a “Signal too quiet to normalize” message appears. The normalized version is what gets saved when you press Enter to keep the take. If you turn normalization off before keeping, the original levels are preserved.

​
Accessing recordings

If you need to load a previous recording, access an empty channel and press R for Recordings. The file browser will load at your recordings path; To see where that path is on your platform, see here.

​
Failed takes and saving again

Stopping capture drains audio already accepted by the input callback, including its final partial block. A stopped take reaches Review only after capture stops cleanly. If input delivery overflows, Meiji Sampler reports Input capture overflowed; recording discarded, returns to Idle, and offers no truncated take to keep. Existing assigned samples and saved assets remain intact. Re-arm to try again. A stop timeout also discards the take. Input stream failed requires selecting or reinitializing the input device before recording again. If writing a valid reviewed take fails, Review retains the samples so you can correct the storage problem and retry Keep. On MPC Sample, [SHIFT+SAMPLE] selects Microphone or Rear TRS through the native analog controls. [B3] requests application; the applied value changes only on success. [B2] in recording opens the PCM device/channel picker. Input monitoring is off when the recording view opens. Press [M] on desktop or [SHIFT+B2] on MPC Sample to hear the selected input before it is written to a recording. Close the recording view or press the shortcut again to stop it. Input and output must use the same sample rate. On MPC Sample, select an analog source first. Microphone monitoring requires a detected headphone or lineout connection, and stops if that detection is lost. The speaker remains disabled while the microphone source is selected. Physical routing and listening acceptance remain unverified. The monitor queue holds at most 1,024 stereo frames. When independent input and output clocks drift, an empty queue produces silence and a full queue drops new frames. This bounds buffering but does not measure end-to-end latency. On MPC Sample, [SHIFT+B3] in recording enables or disables the USB audio bridge for the current session. WAIT indicates that startup is still in progress; press the shortcut again to cancel. A previous stream must finish stopping before a new one can start. After enabling, choose UAC2_Gadget with [B2] to record host-to-device audio. The post-effects master also goes to the host while analog playback continues. Input monitoring is outside that master return, so it does not feed the same USB output. The bridge requires a unique configured gadget, verified stereo 44.1 kHz descriptors and compatible input and output PCM formats. It refuses a configuration with a mass-storage function. Unknown or disconnected devices leave it unavailable. Connection health is checked periodically; this is not an end-to-end latency guarantee. A disconnect during capture discards the incomplete take; a take already in Review remains there but cannot be saved if its USB source becomes unverified. USB routing and latency require physical acceptance. See native input controls.