Mastering

Voice Mode

Say what you want changed instead of typing it. The engineer hears you, renders the revision, and answers out loud in its own voice.

Press the microphone, say it, press Enter. The answer comes back spoken.

Voice Mode is Prompt to Mastering with your mouth instead of your keyboard. Everything a typed prompt can do, a spoken one can do too, and the engineer talks back.

#Starting a conversation

The microphone lives in the prompt bar, right before Apply Prompt.
  1. Press the microphone
    Playback pauses and the Voice Mode window opens. The first time, your browser asks for microphone permission; say yes.
  2. Say what you want
    When the status reads Listening, talk. Your words appear in the box as you speak. Anything the transcription got wrong, just fix by typing in the same box.
  3. Press Enter
    Or click Apply Prompt. The window closes, the prompt goes to your engineer exactly like a typed one, and a new revision starts rendering.
  4. Listen to the answer
    When the revision lands, the engineer reads its reply out loud in its own voice. The text is in the chat as well, so nothing is lost if you had the volume down.
Listening. The transcript lands in the box while you talk, and Enter sends it.

#What to say

Talk the way you would to an engineer in the room. Short and concrete works best, and two wishes in one breath are fine.

  • "Warmer, and bring the vocal forward a little."
  • "Too much low end, tighten the bass and keep the punch."
  • "Make it sound like a late 90s rock record, but not louder."

Speech recognition is multilingual, so use whatever language you think in. Words like warmer, brighter, wider and punchier are read the same way as on the Prompt page.

#Who is talking back

Each mastering engineer has its own voice. Switch from Default to Sophia and the reply changes character along with the sound. Secret Fun Mode changes the delivery as well; yes, Angry sounds angry.

  • Talk or type while the reply is being read and the readout stops; you are more important than the speech.
  • The reply keeps playing with the window closed. Press the microphone again if you want silence.
  • The reply is the same text you see in the chat, so re-reading it later is a scroll away.

#The status line

The line at the top of the window tells you what is going on. Open the window again while a revision renders and it picks up where things are.

StatusMeaning
Connecting microphoneAsking for the mic and opening the transcription link. A second or two.
ListeningTalk. Interim words show up in grey under the box, final ones move into it.
Sending promptYour words are on their way to the engineer.
Waiting for responseThe revision is rendering. Same wait as a typed prompt.
Reading responseThe engineer is speaking. Press the microphone or type in the prompt bar and it stops.
Response readDone. Press the microphone to say the next thing.
Needs attentionSomething failed, usually microphone permission or a dropped connection. The message under the box says which.

#Good to know

  • A voice prompt is a normal revision: it counts like any other one and shows up in the Mastering A/B list with its transcript as the name.
  • Prompts can be up to 500 characters and need at least 5 to send. If nothing is heard for a minute, listening stops on its own; open the window again to continue.
  • Opening Voice Mode switches the bar to Prompt mode. Reference mode has no voice, the reference track already says it all.
  • Audio is streamed for transcription only while the window is listening and is not stored. The transcript is kept as the revision's prompt, like a typed one.
  • Needs a browser that can record from a microphone: current Chrome, Edge, Safari or Firefox. Public share links cannot start revisions, so no Voice Mode there either.
Does Voice Mode cost extra?

No. A spoken prompt makes a normal revision and follows the same revision limits as typing.

Can I use it in the mobile app?

Not yet. Voice Mode is web only for now; the app has typed prompts.

It says Needs attention as soon as it opens.

Check the microphone permission for mastering.soundboost.ai in your browser's site settings, then press the microphone again. On a company network, a firewall that blocks websockets shows up the same way.

Can I edit what it heard?

Yes. The box is a normal text field. Fix a word, add a sentence, then press Enter. When you close the window, whatever is in the box is copied into the prompt bar.

Ready to hear it on your track?

Upload a mix, get the first master in about a minute, and keep the revisions you like.

Try it on your own track

The first master takes about a minute. Previews are free.

Open Mastering