This Week the Tools Were Told to Leave the Singer In

A haunted doll from the 1890s, made by Thomas Edison, turned up on the wire this week. CDM filed it in a piece about the oldest recorded voices, haunted dolls and more from Tempest Instruments, and admitted the doll got their attention faster than any new sound technology. I understand why. A recorded voice coming out of a small machine is the oldest trick in our trade, and the unease is the same one it always was: how much of the person is still in there.

Almost every other vocal story on the wire this week was an argument about that question, and the loudest came from Antares. Eleven items on our wire carried AutoTune Advanced, in English, French and Spanish, and the line that MusicRadar put in its headline is the one I would have chosen: since 1997, the complaint about pitch correction has been that it takes the singer out of the singing. It is rare to see the old complaint against a product sitting in the headline of its own launch. Sound On Sound describes the plug-in as a ground-up rebuild of the Auto-Mode algorithm with a new DVA vibrato-detection engine.

One mechanism, once. Vibrato is a singer moving the pitch a little above and below a note, several times a second, on purpose. An automatic corrector that pulls every moment toward the nearest note hears that movement as an error and flattens it, and a flattened held note is the sound people have been complaining about for almost thirty years. Audiofanzine gives the cleanest summary of the fix: the engine is there to protect the natural vibrato of a voice while it corrects. Detect the wobble, leave it alone, tune the note underneath it.

The rest of the sheet is for engineers. MMR reports sub-1ms latency and a feature set kept separate from AutoTune Pro 11, and Hispasonic adds a real-time design with new formant manipulation. Latency is the gap between the singer making a sound and hearing it back; under a millisecond is small enough that a singer tracking through the plug-in should not feel pushed. Production Expert ran a review under the question what makes it different, and marks it as partner content, which you should know before you read it. I have not put a voice through AutoTune Advanced yet, so I will not tell you whether the vibrato survives. I will tell you the goal is the right one.

The second tool of the week works on the same problem from the other side of the session. Synchro Arts released VocAlign 7, and Production Expert's headline says what changed: you can now align entire vocal stacks in one pass, in a multitrack workspace. DTM Station in Tokyo explains the idea for anyone who has never used it. The lead vocal is the guide, and the tool brings the timing and pitch of the doubles and harmonies into line with it; their headline says it cuts vocal editing time to a tenth. I spent a lot of late nights as an assistant in Los Angeles nudging background vocals by hand, syllable by syllable, and I would have given a week's pay for that tenth. What I like about the design is where the authority sits. The machine does not decide when a word lands. The lead singer already did, and the rest of the stack is asked to follow.

Then the front of the chain. Neumann launched the KK 104 A and KK 105 A, condenser capsule heads for specific Sennheiser wireless handheld transmitters, which MMR says are meant to enhance vocal performance. The capsule is the part of a handheld microphone that actually hears; the body under it is mostly the radio. Swapping one capsule for another is a decision about what the stage gets of a voice before any software touches it. It is also the cheapest place to leave the singer in.

The strangest vocal story was about paperwork, and it was the one that sent me back to Edison's doll. Digital Music News reports that Voice-Swap has joined the Digital Data Exchange, DDEX, as a full member, with an eye on consent and authorization metadata for AI voice models, because AI vocals have become, in the company's words, part of real production and commercial workflows. A tuning plug-in asks how much of the singer is left in the note. A metadata field asks whether the singer said yes. The second question is newer, and it is going to sit in the delivery spec of records I mix whether I like it or not.

And a record, because the tools are for records. Magnetic writes that Medicine's "Just Walk Away", out on Make Your Era on 25 September, works because it knows what needs to stay in the center, with the rolling drum pressure and low-end movement around it. Vocal first. That is the whole brief for every tool above, written by a reviewer about a drum and bass single.

There is a session open in my room in Sangenjaya tonight, a lead and a stack of doubles, and on the last word of the chorus the singer drifts a quarter of a semitone down and comes back. It is not in tune. It is her. The cursor is sitting just to the left of it, and I have not moved it.