# Step 6 · Audio

This step sets up the therapeutic sounds and the rules that match the right sound to
each participant. It has several parts, worked through in order.

## The building blocks

Hearing profiles
: The presets that describe a participant's hearing (for example None, Mild, Moderate,
  Severe). A participant is matched to one of these.

Stimulus types
: The kinds of stimulus used in tinnitus matching.

Modulation bands
: The frequency bands a sound can be modulated in (for example 2–4 kHz).

Frequency mapping
: The table that maps a participant's matched tinnitus frequency to the modulation band
  they should receive, with separate active and placebo columns so sham arms get the
  control band.

## Uploading the sounds

Two kinds of audio are uploaded:

- **Matching audio**: used during tinnitus matching.
- **Therapeutic sounds**: what participants listen to during the trial.

Each therapeutic sound is identified by three things: its **sound type**, the **hearing
profile** it is for, and its **modulation band**. Files are recognised by name using the
convention:

```
{SoundType}_{HearingProfile}_{ModBand}
```

for example `Amp_Mild_2-4`. Unmodulated sounds (used by *washout (unmodulated)* arms)
have no band.

The step shows a coverage summary per sound type, how many of the expected
hearing-profile × modulation-band combinations you have uploaded, so you can see at a
glance what is still missing.

## How a sound is chosen for a participant

When a participant is ready to listen, the portal selects their sound from the sound
type of their current arm, their confirmed hearing profile, and the modulation band
their matched tinnitus frequency maps to (active or placebo, depending on the arm). The
full matching logic is described in {doc}`../audio-pipeline`.

Next: **{doc}`7-compliance`**.