A singer who hears the tuned voice sings to it. Getting there without a late headphone mix means buffer, plugin mode and monitoring set in the right order.
Hard tuned vocals are not only a mix decision. A rapper or singer who hears the stepped, locked sound in their headphones performs differently: they lean into the notes the effect lands on, hold vowels where the snap sounds good and stop fighting pitches the correction will move anyway. Recording dry and tuning later gives you a cleaner file and a different performance. The catch is time. Any pitch correction needs a moment to hear a note before it can move it, and if that moment plus your audio buffer reaches the headphones late, the singer stops singing in time.
Two things decide how late the tuned voice arrives. The first is the audio interface buffer: the larger it is, the longer every sample waits on the way in and out. The second is the plugin itself. A tuner that looks further ahead can track pitch more cleanly, and that look ahead is latency the DAW has to account for.
When the plugin reports its latency, the DAW lines the recorded audio up correctly afterwards, so the take is never out of time on the timeline. What it cannot fix is the headphone mix while you record. A singer can tolerate a few milliseconds before the voice feels detached. Tens of milliseconds sound like a slap back echo of their own voice, and the performance suffers. So the job is to keep both numbers small while tracking and let them grow back once the take is done.
Pick the key and scale of the song first. Correction that does not know the key snaps to the nearest semitone, which is chromatic correction, and in a hard setting that sounds like wrong notes. If the song changes key, note where, and either automate it or record the sections separately.
Lower the interface buffer as far as the computer allows without clicks while you track. Freeze or disable heavy plugins on other tracks if you need the headroom. This is a tracking setting, not a mixing one, and you raise it again later.
Use the plugin's tracking or low latency mode while recording. It trades some tracking accuracy for speed, which is the right trade when the output is only going to headphones. Choose the input type that matches the voice, because a low male voice and a soprano are tracked differently.
If the interface also has direct monitoring on, the singer hears the dry voice early and the tuned voice late, and the two together sound worse than either. Use the software path only, or blend in a small amount of dry signal deliberately.
Put the tuner on the monitoring path and record the input clean. The singer performs to the effect, and you keep the option of changing the retune speed or the key afterwards. Before the mix, switch the plugin back to its full quality mode and raise the buffer.
TUNELOCK is built around exactly this split. Its Low Latency mode brings the delay to about 2 ms so the singer hears the tuned voice while recording, and its full quality mode, about 56 ms at 48 kHz, reports its latency to the DAW for the mix. RETUNE SPEED goes from a glide nobody hears to zero for the hard, stepped sound, FORMANT keeps the voice its own while the pitch moves, and the input type covers soprano, alto tenor, low male, instrument and bass. It is made for a clean single lead vocal rather than a stack or a choir. If the room is the bigger problem in your recordings, start with recording vocals in an untreated room, and keep the tuner early in the vocal chain order.
TUNELOCK runs at about 2 ms in Low Latency mode so the singer hears the tuned voice while recording, then switches to its full quality mode for the mix.
Yes. Put the pitch correction on the monitoring path in its low latency mode, keep the interface buffer small, and record the dry input so the tuning can still be changed later.
The interface buffer and the plugin's own look ahead add up. Lower the buffer while tracking and switch the tuner to its low latency mode, then switch back for the mix.
Set the key and scale of the song and turn the retune speed to zero, so every note snaps to the nearest note in the key with no glide.
Guides
You may also like
Sound DesignVocalsPitchPitch is the note, formant is the size of the voice singing it. Moving them separately gives bigger, smaller or doubled vocals with the melody untouched.
VocalsRecordingReflections are a distance problem, not a gear problem. What to change at the microphone in an untreated room, and which damage no later pass can undo.
