TL;DR. How does DAF work? Delayed auditory feedback inserts a 50–200 ms delay between speaking and hearing your own voice. The delayed signal may give the speaker an external timing cue, which can support slower, softer speech. The mechanism is related to the choral speech effect, but long-term carryover varies.
The basic mechanism
Speech is a closed-loop motor task. When you speak, your brain compares what you intended to say with what you hear yourself say, and feeds that comparison back into the motor control circuits to adjust in real time.
People who stutter appear to have subtle differences in how this feedback loop is timed. Research discusses basal-ganglia–cortical timing circuits and auditory-motor integration as part of fluent speech production, and these systems may handle the relationship between motor command and auditory result differently in people who stutter.
DAF may use this loop in a helpful way. By imposing a deliberate delay (often around 100 ms), it gives the auditory system a clear external cue to pace with — which may push speech toward a tracking mode that resembles part of choral speech.
Two consequences:
- Speech rate slows. The speaker waits, often unconsciously, for the echo before producing the next sound.
- Articulation softens. Onsets are gentler. Blocks become less common.
The choral speech effect — the closest analogue
Read aloud in unison with another person, and many people who stutter become much more fluent — including some with severe baseline stuttering. This tells us something important: the speech-motor system in people who stutter can often produce fluent speech under the right conditions.
DAF approximates part of choral speech with one speaker. The delayed playback acts as a voice-like cue. Frequency-altered feedback (FAF — pitch-shifted playback) may do something similar through a different route. Combined feedback (DAF + FAF) is sometimes used in clinical and device settings.
What changes inside the brain
Neuroimaging studies of stuttering under altered-feedback and choral conditions report patterns such as:
- Increased activity in right hemisphere speech regions — the right temporal cortex in particular.
- Reduced abnormal activity patterns in basal ganglia circuits.
- Reduced compensatory activity in motor planning regions.
- A left-right hemispheric balance that, in some studies, looks closer to fluent-speech patterns during the task.
In short: DAF appears to recruit auditory-tracking and compensatory networks that may reduce the impact of stuttering-related timing differences during use. This is a useful working model, not a fully settled explanation.
Choosing the delay length
The delay number is the time between producing a sound and hearing it back. At 100 ms you hear yourself a tenth of a second after you speak — long enough to register as a separate voice, short enough not to feel like a chaotic echo.
The functional range is roughly:
- 0–30 ms. Below the perceptual threshold for most people. Speech feels normal; the delay is unnoticed; effects on stuttering are minimal.
- 50 ms. Subtle delay, perceptible. Light effect. Used by experienced speakers after months of practice.
- 75–100 ms. The mainstream range. Strong effect, manageable speech rate.
- 150 ms. Heavy effect. Speech feels markedly slowed; useful in therapy or on hard days.
- 200 ms. Maximum useful delay. Speech is very slow; mostly used for clinical demonstration.
- >200 ms. Speech becomes uncomfortable for most users; effects don't continue to grow.
The principle is the smallest delay that still helps. Bigger delays produce more dramatic effects but are also more artificial; smaller delays are more easily transferred to natural, unaided speech.
Why DAF helps stuttering but disrupts fluent speakers
Bernard Lee discovered DAF in 1951 by accident, working on aviation communication. He noticed that pilots wearing communication headsets, which fed their own voice back to them with a slight delay, started to stumble over their words. The effect was named the Lee effect for a while — and when applied to people who don't stutter, DAF reliably degrades fluency.
The opposite happens in people who stutter. Why?
One practical hypothesis: in people who do not stutter, live auditory feedback is already well matched to speech-motor output, so inserting a delay creates a confusing echo. In many people who stutter, a short external delay may instead give the auditory system a clearer timing cue to follow.
This is the simplest mechanistic story that fits the data. Competing models exist — including the EXPLAN theory and various motor-timing accounts — but for practical purposes, the choral-speech analogy is the most useful way to picture it.
Carryover — the unsolved part
The biggest open question with DAF: does the benefit carry over to non-DAF speech?
The honest answer is partially, and it varies between people. Many adults find that DAF practice over weeks to months improves their non-DAF speech — they learn the slowed pacing and gentler onsets and keep producing those patterns when the headphones come off. Others find DAF helpful only while it's active. Which outcome you get isn't fully predictable.
The practical move is to use DAF as a training tool: 5–10 minutes of practice with DAF, then 2–3 minutes of transfer practice without it. The transfer block is what builds carryover.
Why hardware vs app doesn't matter much for the mechanism
DAF is just an audio-processing function. Whether it lives in an in-ear hardware device or a smartphone app, the core signal path is the same: capture voice, delay it, play it back through headphones. What differs between products is form factor, latency stability, extra features (FAF, masking), and ergonomics.
The one important caveat is wireless headphone latency. Bluetooth adds its own variable delay (often 150–300 ms) on top of whatever the DAF software adds. That extra latency is unstable and undermines the precision the technique depends on. Use wired earbuds for any serious DAF practice.
Where to learn more
- What is DAF — the basics and why it helps.
- The science of DAF — the research base and effect sizes.
- Stuttering treatment — where DAF fits among other approaches.
- Exercises — daily practice routines that pair with DAF.
Practise for 5-15 minutes with wired headphones. StutterFlow is a practice tool, not a cure or a replacement for speech therapy.
