Background noise and room echo

Two settings in the player's Background noise and room echo menu place the sound somewhere. Background noise adds the chatter of a busy restaurant, and Room adds the echo of an everyday room. Both join the sound before the hearing loss, so every profile, hearing device, and export includes them, and A/B keeps them. Together, they let you hear a conversation across a restaurant table: other people talking around the voice you're following, and the room's echo on it.

Background noise

Other people talking cover the quieter parts of the voice you're following. With sensorineural hearing loss, following speech in noise usually takes speech that stands further above the noise than it does with normal hearing.¹

Speech to noise

The slider sets the speech-to-noise ratio, from −5 to +20 dB: how much louder the speech samples are than the chatter. At 0 dB they're equally loud, and at +20 dB the chatter sits far below the speech. It starts at +5 dB. The level is set against the two speech samples, so with another source, such as a file you upload or your microphone, the slider still sets the chatter's level relative to the speech samples, not to that source.

The recording

The background noise is Chattering in a restaurant, recorded on a phone by Joshun and shared on Freesound under CC0. The simulator plays 40 seconds from the middle of the three-minute recording, the steadiest stretch, with no loud clatter, cleaned of room rumble below 90 Hz. It loops without a seam. The recording is mono, so the right ear hears it from 20 seconds further on, and the room surrounds you rather than sitting in the middle of your head.

Room echo

In a room, a voice reaches you twice: once directly, and again as echo off the walls, floor, ceiling, and furniture. The echo rings on after each sound, filling the short gaps between words and covering quiet consonants, much as other voices do. With sensorineural hearing loss, telling the voice apart from its echo takes more work, so a room with hard surfaces can make speech hard to follow even when it's quiet. If you have hearing loss, you may already know which rooms those are.

The Room setting plays any source as if it were heard from across a table, about 1.5 m away, in one of six kinds of everyday room. The background noise doesn't get the room's echo, since the chatter was recorded in a room of its own.

The rooms

The rooms are listed from least echo to most. Echo time is how long the echo takes to die away, by 60 dB, which is the figure the simulator shows. Late echo is how loud the echo is from 20 ms after the voice arrives, compared with the voice itself, between 500 Hz and 4 kHz.

Echo time alone doesn't decide how echoey a room sounds. A kitchen's echo dies away sooner than a classroom's, but it starts louder, so it covers more of the voice.

Where the rooms come from

James Traer and Josh McDermott at MIT measured how 271 everyday places echo, chosen by texting volunteers at random times over two weeks to ask where they were.² Their speaker and microphone were 1.5 m apart, the distance the simulator uses. Each room here is the median of the places of that kind in their survey. Only those measurements are used, not the recordings.

The echo itself is made in your browser the way the paper made its own: the direct sound, then noise that fades at each pitch at the rates the survey found. Echo lasts longest from about 200 Hz to 2 kHz and dies away faster above and below, more so in echoier rooms, and it starts quieter at the highest pitches. When the researchers played echo made this way alongside recordings of real rooms, their listeners couldn't tell which was which.

Each ear hears its own echo, so the room surrounds you rather than sitting in the middle of your head. A room doesn't make the sound louder overall. It spreads the same sound out over time.