The sound desk feed

Yazyk translates a spoken feed and sends one translated language to an assistive-listening transmitter. Somebody else runs the computer; this page is about the route and the levels at the desk.

The signal path in

pulpit or lapel mic -> desk direct out, aux, or mix-minus -> USB interface -> Yazyk

The best feed is the pulpit or lapel microphone on its own. A direct out, an aux send, or a mix-minus all beat the main LR mix, because singing is still a human voice: if songs reach Yazyk, they can be translated and spoken over the top of the room.

This is advice and not a requirement. If the desk can only give you the main LR mix, send that and say so. The software has to handle that case and does.

Do not route Yazyk's translated output back into its own input.

Input level

A median of about -20 dBFS on normal speech, with no clipping, is a healthy reference rather than a number to hit exactly.

The output to the transmitter

USB interface line out -> one analog cable -> transmitter input -> bodypack receivers

• Use the interface's line output, and set the transmitter input to line level if it offers that.
• Start the output path low. Raise it only while listening on a real bodypack receiver.
• If the transmitter expects mic level, use its pad or a suitable external pad. A DI box helps if the rack and the computer together produce hum.
• One cable carries one language. A second transmitter language needs a second output path.

Do not set the level by the digital peak. A translated output can reach full scale and still be painful in the room, and transmitters have been reported as too loud even at their lowest setting. Set the final level by ear, at the back of the room, before people arrive.

Who owns the USB input

Many USB interfaces are exclusive-access: only one program on the computer may hold the input at a time. That single fact causes more Sunday morning failures than anything else at the desk.

• Do not select that input in video-mixing, meeting, recording, or browser software while Yazyk is using it.
• If another program genuinely needs the same audio, give it a different input, or prove the shared arrangement at a midweek rehearsal rather than improvising on the day.
• If Yazyk fails to start just after a stop, wait a few seconds and start once more. The interface may still be releasing the input. It is not evidence of a bad cable.

The operator presses Start once for the whole service and uses Input Mute between speaking segments, which keeps the interface held open the whole time. Output Mute is reserved for silencing an over-loud output immediately.

Write these down for next time

Filling that in once, at a rehearsal, is what stops the next volunteer rediscovering it under time pressure.