Voice

Voice lets you talk in a session instead of typing, and listen instead of reading. It's useful when you're thinking out loud, away from your keyboard, or on a phone. You speak into the microphone in the message box and your words appear as you talk; you can also have new messages read aloud as they arrive. Voice is on for everyone in an organization unless an Owner or Admin switches it off for someone.

Send a voice message

In any session, click the microphone in the message box (Voice input) and start talking. Your words appear as a live transcript while you speak, so you can see what DevSpec heard before you send it.

When you've finished:

While you're recording, the picker next to Send lets you change where the message goes. Recording carries on until you send or cancel; it doesn't stop when you pause. A recording can last up to 15 minutes, so send long thoughts in parts.

A voice message goes to the same place a typed one would. A new session's Send to starts on Post to room, so pick Dev first if you want Dev to answer what you say. See Sessions.

Speech-to-text understands English only. To choose which microphone to use, open the composer's settings (the gear beside the message box) and pick one under Microphone.

The same microphone works in Messages, your one-to-one conversations, for sending voice notes. See Notifications and messages.

Transcribe a saved recording

Choose an audio file with the paperclip in a session. DevSpec adds its transcript to your draft so you can check and edit it before sending. The recording is not attached to the message as a playable file.

Recordings can be up to 15 minutes and 50 MB. Transcription is charged as Voice input, just like dictating with the microphone.

Answer a question by voice

When Dev or an agent asks you a question in a card, choose Listen to hear the question and its choices. Use the microphone to dictate a written answer or your own alternative. Submit finishes an active recording and sends your answer.

Have messages read aloud

You choose what happens when a new message arrives in a session you have open. It's your own setting, the same in every session, and it doesn't affect anyone else. Set it in Settings → Replies & voice → Playback under New-message sound, or in the composer's settings (the gear beside the message box) under Voice output:

• Silent (the default): nothing plays automatically.
• Beep: a short chime when a message arrives.
• Voice: new messages are read aloud.

In Voice mode, new messages from Dev, teammates and agents are read aloud. Your own messages never are.

Any message can also be played on demand: press the speaker control under it. In Voice mode, playing one message carries on through the messages after it. Playback speed (0.75x, 1x, 1.25x or 1.5x) is in the same two places.

If someone chose to keep their voice recording, playing their message plays their real voice. Otherwise the words are read in your chosen voice. Very long messages can't be read aloud ("Message is too long to read aloud.").

Choose a voice and language

In Settings → Replies & voice:

• Voice sets how DevSpec sounds when it reads messages to you. There are 28 voices, grouped Female and Male; the default is Kepler. Type something into Preview text and press Play preview to hear a voice before you choose it.
• Voice language sets the language messages are read aloud in. Choose the language your conversations happen in: English, Spanish, French, German, Portuguese, Chinese, Japanese, Korean, Hindi, Arabic, Italian or Russian. It doesn't translate messages, and it doesn't change what you can speak in: speech-to-text is English only.

Both are personal. Each person hears messages in the voice and language they chose. See Your settings.

Let teammates hear your real voice

When you speak a message, DevSpec always keeps the written words. Whether it also attaches your voice audio is up to you, and it's off by default.

Switch it on with Save my voice in messages in the composer's settings (the gear beside the message box) or in Settings → Privacy & data. They're the same setting.

• On: people in the session can play your real voice on the messages you speak.
• Off: only the words are kept, and anyone who plays your message hears it read in their own chosen voice.

Voice notes you send in Messages always include your voice, whatever this setting says. All voice recordings are kept for 48 hours and then automatically deleted—only the written message stays. See Your settings.

Start a session by voice on your phone

On a phone, a Start a new session button floats at the bottom of every project page, except inside a session, where the message box has its own microphone. Tap it to start a new session by voice (Talk) or text (Type).

First choose who receives what you say:

• Start a session without an agent (the default): create the session with what you said, and invite others later.
• Dev: create the session and ask Dev to respond.
• One of your agents: send it to an agent you have running.

Then speak in Talk mode, or switch to Type, and submit. DevSpec creates the session and opens it.

On a phone, the New session button on the project Overview is hidden, because this button does the same job. See Project Overview.

You can choose who can see the new session before creating it: the project team, only you, or specific people. See Sessions.

Who can use voice

Reading aloud and voice input are on for every member of an organization by default. An Owner or Admin can switch either off for a person in Settings → Usage & limits, in the Reading aloud and Voice input columns under People. Each switch stops only its own half: with Voice input off the microphone stops and replies can still be read aloud, and with Reading aloud off playback stops and the person can still dictate. DevSpec tells them which is off and that an admin of the organization can turn it back on.

Reading aloud and voice input can also stop because of a limit or a pause. DevSpec tells you which one:

• your personal allowance for the period, which Dev, reading aloud and voice input share
• the project's own spending limit
• the project being paused by an Owner or Admin
• the organization's balance running out

The recording or reading that reaches a limit finishes and is charged in full; the next one is what stops.

Owners and Admins change these limits, and pause or resume projects, in Settings → Usage & limits. See Billing, usage and limits.

If voice isn't working

What you might see, and what to do:

• Voice not supported on the microphone: this browser can't record audio. Update it, or try another browser.
• Microphone permission denied: allow microphone access for DevSpec in your browser's site settings, then try again. The composer's settings also offer Grant access under Microphone.
• No microphone found: connect a microphone, or choose a different one under Microphone in the composer's settings.
• Microphone is busy: another app or tab is using it. Close it and try again.
• No speech detected. Please try again.: DevSpec didn't hear anything. Check the right microphone is selected and speak a little closer.
• The connection drops while you're talking: DevSpec reconnects by itself a few times. If it can't, it puts everything it heard so far into the message box, so you can check it and send it.
• A message that an admin can turn reading aloud or voice input on, or that a limit has been reached: one of them is switched off for you, a limit stops it, or the project is paused. Ask an organization Owner or Admin, who can change all of these in Settings → Usage & limits.
• Message is too long to read aloud.: read that message instead, or play a shorter one.

What voice costs

Voice is charged to your organization's balance, in dollars: voice input for the time you speak, and reading aloud for the length of what's read. Charges are listed as Voice input and Reading aloud in Settings → Billing → Transactions.

Speech-to-text is charged whoever the message goes to, including a message posted to the room that Dev doesn't answer.

Reading aloud and voice input count toward each person's allowance, which your Owners and Admins set in Settings → Usage & limits, alongside Dev's replies. Voice mode reads every new message aloud, so it costs more than playing only the messages you choose. Beep and Silent cost nothing. See Billing, usage and limits.