CI Atlas · Hearing Music Through an Implant · Module 03

3What the Implant Keeps and What It Throws Away

Envelope coding preserves the rhythm of music and discards almost everything else: this trade-off is the engine of the whole chapter.

FEnvelope coding keeps the rhythm

Modern strategies such as CIS and ACE extract the slowly varying amplitude envelope from each frequency band and use it to modulate fixed-rate electrical pulses on the corresponding electrode. Because rhythm and tempo are themselves slow envelope-level patterns, they pass through this coding almost intact, which is why rhythm is the best-preserved musical dimension in implant users. Studies confirm that implant recipients discriminate tempo nearly as well as normal-hearing listeners, yet recognise familiar tunes well only when a memorable rhythmic line is present. When rhythm cues are stripped out and only pitch remains, melody recognition collapses toward chance, exposing how heavily implant users lean on the surviving temporal frame.[2004][2004][2020]

Picture a single note as a fast wiggle riding inside a slowly changing outline. The slow outline is the envelope — amplitude changes below a few hundred Hz — and it is what CIS- and ACE-style processors extract and deliver. The fast wiggle inside is the temporal fine structure, the rapid carrier that encodes pitch and timbre, and it is exactly what current strategies discard. Envelope alone is enough for speech but strips the cues music needs. Toggle the layers to see what reaches the nerve. Illustrative.

TFine structure and spectral detail are thrown away

Envelope extraction deliberately discards the temporal fine structure, the rapid waveform detail within each band that in normal hearing carries much of the cue to pitch. Spectral detail is also coarse: although arrays carry 12 to 22 physical electrodes, overlapping electrical fields mean recipients behave as if they have only about four to eight effectively independent channels. This electrode-channel mismatch arises from current spread, where stimulation from one contact excites neurons intended for its neighbours, smearing the spectral picture the brain receives. With pitch's temporal cue removed and its place cue blurred, melody and timbre, which need fine pitch and clean spectra, are the casualties, while the rugged rhythmic frame survives.[2001][2008][2004]

A modern array holds 12–22 physical electrodes, which sounds like fine spectral resolution. But because electrical current spreads through the conductive perilymph, adjacent electrodes excite overlapping populations of nerve fibres, so the number of channels a listener can actually use independently falls to roughly 4–8. Tap a bar to compare. Speech rides comfortably through that bottleneck, but music — needing fine pitch and many simultaneous voices — does not, which is the hardware-level reason music perception lags. Illustrative.

A cochlear implant offers the brain two ways to signal pitch, and both are blunt. Place pitch moves the percept by switching to a different electrode, but there are only a dozen or so active contacts, so the scale climbs in coarse, widely-spaced steps rather than the smooth continuum of a healthy cochlea. Rate pitch raises perceived pitch by stimulating faster, but it saturates near a few hundred pulses per second (about 300 Hz here): push the rate higher and the pitch barely rises. Together these ceilings are why electric hearing carries melody so poorly. Schematic.

CPlace versus temporal pitch in the electric ear

Electric hearing can signal pitch two impoverished ways: place pitch, by choosing which electrode is stimulated, and rate pitch, by changing the pulse rate on a single electrode. Place pitch is limited because evenly spaced electrodes map imperfectly onto the cochlea's tonotopy, and electrically stimulated pitch can differ from the matched acoustic pitch by up to about two octaves. Rate pitch saturates early: raising the stimulation rate raises perceived pitch only up to a few hundred pulses per second, after which listeners stop hearing further change. Both limits leave pitch discrimination coarse, with implant users averaging thresholds around 7.5 semitones against roughly 1.1 in normal hearing, which is why this trade-off shapes every later question in the chapter.[2002][2007][2008]

Case 29.3 · What the Implant Keeps and What It

A recipient with a 22-electrode array tells you he can clap accurately to the beat of any song and follow a drum line, but every melody 'sounds the same' and he cannot tell a guitar from a piano. His engineer asks why so many electrodes give so little melody.

What best explains the gap between his good rhythm and his poor melody and timbre?

Self-assessment — Module 33 questions

Question 1

Which musical dimension survives envelope-based coding (CIS/ACE) best?

Question 2

Although a CI array may have 12-22 electrodes, how many effectively independent channels do recipients typically behave as if they have?

Question 3

Why does raising the stimulation pulse rate on a single electrode fail to keep raising perceived pitch?

Tracked locally in your browser — see /progress for the dashboard.