Show spoken script
Today’s lesson is Unglued masterclass: how to blend a gang vocal in Ableton Live 12 while keeping CPU use low. This is a beginner-friendly, workflow-focused tutorial that shows a lean stock-device chain for stacking, processing and optionally vocoding gang vocals, plus routing, bussing and quick render or freeze tricks so your session stays responsive.
What you’ll build:
- A compact gang-vocal submix made from stacked takes.
- A lightweight vocoded layer using Live’s Vocoder, blended with the dry gang vocal.
- A minimal-CPU processing workflow you can print or freeze to reduce load.
All devices are stock Ableton: Audio and MIDI tracks, Utility, EQ Eight, Glue Compressor, Saturator, Return Reverb and Delay, Vocoder, and Simpler if you want an external carrier.
Overview of the approach:
Stack the vocals, rough timing and pan, route everything to a group bus, add a vocoded layer using an internal carrier for the lowest CPU cost, shape intelligibility and space, then print or freeze the result.
Step A — Prepare your gang takes and submix:
1. Import or record three to six short gang vocal takes on separate audio tracks. Name them logically, like Vox_L1, Vox_L2, Vox_R1.
2. Group them into one track called GangBus. Select the vocal tracks, right-click and Group Tracks.
3. On each individual track, put a Utility device to set stereo width slightly narrower, around 70 percent, and apply small pan offsets to taste.
4. Nudge timing with clip start points or transient markers so the group sits tight. Slight offsets make the stack sound bigger.
Step B — Clean and glue the submix with minimal processing:
1. On the GangBus, insert EQ Eight first. High-pass around 110 to 140 hertz to remove rumble. If it’s muddy, make a small dip around 300 to 500 hertz. Add a small presence boost around 1.5 to 3.5 kilohertz if words need clarity, plus one or two dB.
2. Insert Saturator set to Soft Clip and keep Drive low — about one to three dB — to add cohesion.
3. Add the Glue Compressor with mild settings: ratio between two to four to one, medium attack, medium-fast release, and aim for two to four dB of gain reduction for glue.
4. Keep the device count light—this gives you cohesion before adding heavier CPU effects.
Step C — Set up the Vocoder:
A quick note: the vocal is the modulator. The carrier shapes the harmonic content. For minimal CPU, use the Vocoder’s internal carrier or a single-voice Simpler if you need character.
Option one — Internal carrier (recommended for CPU savings):
1. Duplicate the GangBus to create a processing layer for the vocoder. Rename it GangVox_Voc.
2. On GangVox_Voc, place EQ Eight before anything else. High-pass at about 120 hertz and tame sibilance if needed.
3. Insert Live’s Vocoder. Set Bands to a moderate value, 12 to 16 bands is a good compromise. Fewer bands use less CPU and emphasize clarity.
4. Set Dry/Wet starting around 40 percent so the vocoded texture blends rather than replaces the dry signal.
5. Keep Carrier on Internal, and set Carrier Pitch to sit with your song’s key, or leave tracking default. Use a short attack and release around 60 to 150 milliseconds depending on vocal speed.
6. Adjust Dry/Wet and attack/release until consonants remain clear.
Option two — External carrier (when you need a specific timbre):
1. Create a new MIDI track, load Simpler, and put in a single-cycle saw or short tone. Play a single sustained note with polyphony set to one.
2. Route Simpler’s output to the Vocoder as the carrier. This uses more CPU than Internal, so print once satisfied.
Step D — Make the vocoded voice intelligible:
1. Pre-Vocoder: on GangVox_Voc use EQ Eight to HP at 110–140 Hz, add a mild presence boost at 1.8 to 3 kHz, and shave harsh sibilance around 5 to 8 kHz or use a de-esser if needed.
2. Vocoder settings for intelligibility: start with 12–16 bands, Dry/Wet 30 to 50 percent, a quick attack, and moderate release. Tune carrier pitch if you’re using an external carrier.
3. Post-Vocoder: use a light EQ to carve space and a Utility to control width or wet level. A small shelf or boost around 3 to 8 kHz can add perceived presence without expensive processing.
4. Parallel approach for CPU efficiency: keep the dry GangBus intact and only add the Vocoder on the duplicated GangVox_Voc track at partial wet. Most clarity stays on the cheap-to-process dry bus.
Step E — Space and FX without spiking CPU:
1. Use return tracks for reverb and delay instead of inserting one on every vocal. Create Return A for reverb and Return B for delay.
2. Keep the reverb small and low-diffusion for clarity, and use low feedback on delays.
3. Send the GangBus and the GangVox_Voc to the returns at different levels — for example, 15 to 20 percent from the dry bus and 8 to 12 percent from the vocoder bus.
4. Avoid stacking multiple reverbs across tracks.
Step F — Lock the result with freeze / flatten workflow:
1. When balance is set, right-click the GangVox_Voc track and Freeze Track. Freezing renders the Vocoder and turns off its CPU usage.
2. Optionally right-click and Flatten to convert to audio and delete the original chain, permanently printing the effect.
3. Keep the dry GangBus unfrozen to keep editing options. You now have a printed vocoded layer to blend with the dry bus.
Common mistakes to avoid:
- Running a vocoder on every take instead of on a submix or duplicate — that multiplies CPU load.
- Using too many vocoder bands, for example 40 or more, which costs CPU with little gain for gang vocals.
- Setting Wet to 100 percent without a dry layer — this often makes speech muddier and less intelligible.
- Using per-track reverbs instead of returns; this quickly clogs CPU.
- Not high-passing before the Vocoder; low-end energy smears the effect and wastes resources.
- Freezing or flattening without saving a backup. Always duplicate or save before flattening.
Pro tips:
- Start with 12 bands and Dry/Wet around 40 percent. Automate Wet for choruses if needed.
- Use Utility to reduce the vocoded track width if it clashes with bass and low mids.
- For a thicker gang, duplicate the vocoded track, detune by plus or minus one to three semitones, pan slightly and narrow each duplicate, then freeze them.
- If consonants need help, put a short, fast compressor on the dry GangBus pre-Vocoder to make transients consistent for the Vocoder.
- Bounce stems of your printed vocoded submix and re-import them. A single audio file is the least CPU-heavy option.
- With an external carrier, use a single sustained note and polyphony of one to reduce CPU.
Mini practice exercise — target 20 to 30 minutes:
1. Load three vocal clips onto separate tracks and group them into GangBus.
2. On GangBus: high-pass at 120 Hz, add light Saturator, Glue Compressor with two to three dB reduction.
3. Duplicate GangBus to GangVox_Voc.
4. On GangVox_Voc: EQ Eight HP 120 Hz, Vocoder Internal Carrier, Bands = 12, Dry/Wet = 40 percent, quick attack, release 100 ms.
5. Add a return reverb and send 12 percent from GangBus and 8 percent from GangVox_Voc.
6. Freeze GangVox_Voc. Check the CPU meter before and after freezing to hear the difference.
7. If satisfied, flatten the frozen track and mix the GangBus with the printed vocoded audio.
Recap:
This masterclass showed you how to build a tight gang vocal submix, use Live’s Vocoder efficiently by choosing an internal carrier or a single-voice Simpler, set Bands and Dry/Wet, shape intelligibility with pre and post EQ, blend using parallel routing and returns, and free up CPU by freezing, flattening or exporting stems. Keep your chain lean, print heavy processing once committed, and use shared returns and internal carriers whenever possible.
Extra coach notes — quick mindset and troubleshooting:
- Treat the vocoded layer as texture, not the whole vocal. Keep intelligibility on the dry submix.
- Start with the fewest active devices while auditioning ideas. Print heavy processing only when you commit to a sound.
- Use the CPU meter while switching devices to find big consumers. If CPU spikes, solo the gang bus and vocoder, increase buffer size in Preferences, and freeze or print heavy devices.
- Avoid Complex/Complex Pro warp modes on many clips—turn Warp off or use light modes when possible.
Routing and bussing tips:
- Use one Group/Bus for all gang takes and route the duplicated Vocoder track from that Group, not from many individual tracks.
- Name and color-code GangBus and GangVox_Voc for clarity.
- Use a single pair of returns for ambience and send from both dry and vocoded buses at different levels.
Resampling vs freezing vs flattening:
- Resampling lets you record the bus to a new audio track using Input set to Resampling — this is non-destructive and low-CPU.
- Freeze is quick and reversible. Flatten is permanent — always duplicate before flattening.
- Export stems if you need stability across machines; re-import the exported audio for the lowest CPU load.
Phase, doubling and vocoder band tradeoffs:
- Check transients and nudge stacked takes to avoid phase cancellation.
- For a thicker sound with minimal CPU, duplicate printed audio, detune by one to three semitones, nudge slightly, narrow width and freeze.
- Lower band counts like 8 to 12 are more robotic and cheaper on CPU—good for aggressive DnB textures. Medium bands balance detail and CPU. High bands are heavy and usually unnecessary for gang vocals.
Final mix-checks:
- Listen at typical DnB levels and check in mono to ensure width doesn’t collapse or cause phase issues.
- Use small automation moves rather than static extremes. Automating vocoder wet, send levels, or width adds life without extra processing.
That’s the Unglued masterclass. Follow the practice exercise, keep devices grouped and lean, print heavy processing when you commit, and you’ll get a wide, aggressive gang vocal that doesn’t kill your CPU.
