Mixing: balance first

Identify the lead role

Decide what the listener should notice first: voice, rhythm, or melody. Choose a short, busy passage and build a level balance without processing. If the main part gets lost, start with levels and arrangement.

The leading role can change within a piece. Mark it phrase by phrase before choosing fixed levels or automation. In an experimental session, SEREIN’s tiny noises may matter more than a dense REGOLITH layer. The louder or more broadband part need not be the compositional center.

Panning makes room

Spreading parts across the stereo field can make them easier to distinguish without boosting high frequencies. After changes, check the center and mono compatibility. Take particular care with stereo wideners: a striking spatial impression can come with lost elements when channels are summed.

Stereo placement helps separation but does not resolve every spectral overlap. Check whether a part retains its role in mono and at a lower level. Phase-based widening can also alter the summed timbre. Compare not only impressive width but the stability of the musical pattern.

EQ should solve a stated problem

Before adding processing, put the problem into words: “the guitar masks the vocal” or “the bass booms on one note.” Listen to the change in the full mix. Do not automatically cut the lows on every track; useful parts of an instrument sometimes live there.

A frequency region can be useful in solo yet obstructive in an ensemble. Start with a small change and listen for the relationship it alters. If cutting removes GRAIN’s recognizable texture or HALO’s body, revisit the arrangement and competing layer’s level. The goal is clarity of intention, not an overlap-free analyzer display.

A compressor is no substitute for balance

Compression changes a signal’s dynamics and envelope. If phrases vary widely in level, try drawing or recording volume automation first. After setting the compressor, compare it with bypass at similar loudness to judge the effect.

Attack and release times affect the sense of impact and motion. Compare sparse strikes, a dense pattern, and decay when setting them. Equal gain-reduction readings do not guarantee equal musical results. Keep an uncompressed version and check whether processing is hiding a problem in the original balance.

Space needs context

Choose one main reverb and send several parts to it. Listen for whether the start of a word and the attack of an instrument remain clear. Decay time and predelay depend on the material; there is no universal value for every song.

Direct sound identifies the source; reflections change its apparent distance and surroundings. A long tail can mask small details. Try varying the send through the phrase, keeping the attack closer and the ending deeper. Then bypass the effect entirely: the spatial treatment should support structure rather than merely fill pauses.

Export is a separate check

Save the master version of the project and export a file with a meaningful name. Listen to the beginning, ending, and transitions in the exported file itself. Check for accidentally cut-off tails, silence instead of music, and overload. Choose format, sample rate, and bit depth to meet the recipient’s requirements, while preserving the source project.

Treat the export as an independent object rather than an extension of the open session. Verify the version, required channels, and complete ending. Compare with the project without double monitoring. Include a short version and purpose note when handing it over, reducing the chance that a working example is mistaken for the final master.

A practical check

Save two similarly loud versions of one passage with one meaningful difference. Write the question before listening. If your preference changes with every switch, revisit the parts’ roles or defer the decision until fresh listening. A finished mix is a sequence of understandable choices, not the largest possible number of active effects.