Dictate adds recognized speech to a draft for you to review and send; Voice sends spoken requests during a conversation and reads Tracy’s replies aloud.
Use Dictate when you want to check your words before sending them. Use Voice when you want a spoken conversation. In either case, check the resulting website before relying on a change.
What is the difference between Dictate and Voice?
Dictate adds recognized speech to a draft for you to review and send; Voice sends spoken requests during a conversation and reads Tracy’s replies aloud.
When “Dictate” is offered beside the message box:
- Select “Dictate” and allow microphone access if your browser asks.
- Speak and check the words appearing in the draft. The language indicator shows the language the service heard when it reports one.
- Select “Done” to finish dictation. Review and correct the draft before sending it.
- Send the message when it says what you intend.
Dictation alone does not submit a request. “Cancel dictation” removes that dictation’s words and restores the draft it started with. Typing or pasting ends dictation and keeps the words already added. Pressing Enter during dictation finishes the last phrase and sends the message if that phrase is confirmed, so use “Done” when you want to review first.
For a spoken conversation, use “Voice” in the “Chat or voice” menu when offered. In a new chat, the audio-wave control beside the dictation microphone also opens Voice. Allow microphone access, then speak your request. Voice sends recognized requests as the conversation proceeds, without a separate review of every draft.
You can keep talking while Tracy works. “Send now” sends words waiting on screen, while “Say it again” discards those waiting words so you can repeat them. These controls do not retract a message already sent.
Use “Chat” to return to the written conversation, or “End voice conversation” to end Voice. The messages sent through Voice remain in the chat.
Who can use speech to ask Tracy for help?
Owners, Editors and Viewers can use the speech controls when offered, but speech does not change their site permissions.
- A Viewer can ask read-only questions and add comments, including dictated comments where a microphone is offered.
- An Editor can ask Tracy to change the site within the Editor role’s permissions.
- An Owner can also perform Owner-only actions, such as managing site roles, within the relevant controls and confirmations.
A spoken request does not give a Viewer editing access or give an Editor Owner permissions. See Team, roles and invitations for the full role rules.
How do I choose a microphone or Tracy’s voice?
The Chat or voice menu offers the available voices and microphones, including System default for the microphone.
Open “Chat or voice”, then choose an offered option under “Tracy’s voice” or “Microphone”. The microphone choice applies to both Dictate and Voice. Your browser may need microphone permission before it can show device names. If a selected device is missing, Tracy can fall back to the default microphone and tell you it is doing so.
Allow microphone access for the Tracy page in your browser. If access is still blocked, check the operating system’s microphone permission for that browser and confirm that your microphone is connected. If the browser cannot record audio, continue by typing in chat.
The speech controls depend on the deployment’s speech configuration and the browser’s ability to record on a secure page. Their absence is not proof that your site or account was deleted. Continue in chat and report a persistent problem through Tracy Discord.
Does interrupting speech stop a site change?
Stopping Tracy’s speech does not by itself stop a change already running or undo anything already saved.
“Stop speaking” stops the spoken output. “Mute the microphone” stops microphone input. Ending Voice or returning to chat ends the voice conversation. These actions are different from asking Tracy to stop its work.
To request that running work stops, use “Stop the work” when offered. Check the result: Tracy can report that it could not stop. Even a successful stop leaves anything already saved in place.
If the outcome is uncertain, inspect the conversation, the website and “Site changes” before repeating the request. Read See what Tracy changed and Undo and backups before attempting recovery.
AI can make mistakes. Not every change can be undone, and external effects cannot be recalled.
Can I approve a plan by speaking?
Review and approve plans on screen; do not rely on spoken agreement to approve a plan or every confirmation.
When Tracy asks you to review a plan or answer a question on screen, switch to the written conversation and read the request. Check the action, its scope and the affected site before choosing an answer. A short spoken acknowledgment is not a substitute for reading the plan.
Some ordinary confirmation questions can accept a spoken answer, but that does not establish permission for every action. If Tracy cannot hear the answer clearly or directs you to the screen, use the displayed question. Never assume that saying yes has approved or completed a site change.
Why did dictation or Voice stop?
Dictation and Voice can stop or pause because of time limits, silence, microphone access, a background tab, another listening session or a connection problem.
Read the displayed reason before restarting:
- A dictation session stops after five minutes, or after 30 seconds without recognized words. Check the draft, then use “Dictate again” when offered to continue.
- If Tracy reports that today’s dictation or voice time is used, continue by typing in chat. The message concerns speech use, not a promise that another allowance remains.
- Dictation finishes when its tab goes into the background. Voice pauses while the tab is in the background. Return to the page and use “Talk again” when offered after checking the conversation.
- If another tab or window takes over listening, use the intended tab and check its current state before starting again.
- If the microphone disconnects, reconnect it or choose a working microphone. If speech settings changed, start Voice again when prompted.
- If the service is busy or the connection is lost, follow the displayed message and continue in chat if needed. If Tracy says your words could not be sent, check the conversation before repeating them.
If Tracy warns that the last words may be missing, read the draft and add any missing words manually before sending. Do not assume that the entire spoken request was recognized.
If this does not work: Send the affected site, time and time zone, browser, displayed error and whether Dictate or Voice failed to Tracy Discord. Remove private information from screenshots and copied details.
What if Tracy cannot read the reply aloud?
If a spoken reply fails or the daily spoken-reply allowance is used, the reply remains in chat.
Open “Chat” to read the reply and any questions or errors. A spoken summary can be shorter than the written response, so also check chat when you need the complete details.
Silence does not prove that the work failed. Check the conversation and the site before sending the same request again. If audio remains silent, check your device’s output volume and browser sound settings, then continue in chat.
Where is my audio processed?
Your microphone audio is sent to a third-party speech service used by Tracy for transcription, and reply text is sent to that service to generate spoken replies. Speech is not processed only on your device. Words you send become part of the conversation; dictated comments become comments when you post them.
Do not speak passwords, access tokens, sign-in codes or unnecessary sensitive information. For provider processing, training, storage location or applicable data terms, contact Tracy Discord before sharing sensitive information.
Read Security and your data for the wider data and access rules.
Frequently asked questions
What is the difference between Dictate and Voice?Dictate adds recognized speech to a draft for you to review and send; Voice sends spoken requests during a conversation and reads Tracy’s replies aloud.
Who can use speech to ask Tracy for help?Owners, Editors and Viewers can use the speech controls when offered, but speech does not change their site permissions.
How do I choose a microphone or Tracy’s voice?The Chat or voice menu offers the available voices and microphones, including System default for the microphone.
Does interrupting speech stop a site change?Stopping Tracy’s speech does not by itself stop a change already running or undo anything already saved.
Can I approve a plan by speaking?Review and approve plans on screen; do not rely on spoken agreement to approve a plan or every confirmation.
Why did dictation or Voice stop?Dictation and Voice can stop or pause because of time limits, silence, microphone access, a background tab, another listening session or a connection problem.
What if Tracy cannot read the reply aloud?If a spoken reply fails or the daily spoken-reply allowance is used, the reply remains in chat.
Where is my audio processed?Your microphone audio is sent to Tracy’s configured speech service for transcription, and reply text is sent to that service to generate spoken replies.