Speech

Have results read out loud to you and launch Agents with your voice

Read Agent responses out loud (Text-to-Speech)

The floating panel and Steve's chat bubbles both have a speaker button on them in the upper right that you can press to have the Agent results read out loud to you.
Pressing the button while reading is in progress will stop it. So will closing the floating panel or the Chat.
You can customize the text-to-speech settings on the Speech Settings page.
Settings that you can modify include: language, voice, speed, pitch, volume and whether Ask Steve should automatically read all agent responses.
NOTE: The Google-branded voices in Chrome are really good, as are the Microsoft-branded ones in Edge.
This should work in all browsers that Ask Steve supports.

This uses the browser's built in text-to-speech capability. Ask Steve doesn't send anything to another server.

"Read this to me" Agent

We added a new "Read this to me" Agent that will take the main content from any page and read it out loud to you, even if you don't have auto-read turned on. And you can trigger it with either a 1-click Agent Button or your voice, so now you can easily have any page read out loud to you. You can install it here.

Launch Agents with your voice (Push-to-Talk)

If you are using Chrome (any platform) or Edge (PC only), you can launch Agents with your voice. Edge on Mac isn't currently working due to a Microsoft issue.
When you hover over the purple lightning bolt tab, a microphone button appears. Press and hold it, say either the name of the Agent you want or make a One-Time Request and then release the button.
If you haven't already, you will be prompted to give Ask Steve microphone access. Press the Give Permission button that appears and then press the Setup Speech Recognition button on the Speech Settings page. Then test if your browser supports recognition in the following section. If it all works, go back and try pressing the microphone button near the purple tab again.
You will see the speech recognition in process in a purple window near the top of the screen.
When you release the button Ask Steve will try to match what you said against one of your Agents, or if there is no good match, it will assume you are making a One-Time Request, and then execute what you asked for.
There is also a microphone button in the Chat window that works the same way.
An even faster way to launch Agents is to use the keyboard hotkey instead of the microphone button. By default the hotkey is Alt-Z (Windows, ChromeOS) and Option-Z (Mac) but you can change this on the Speech Settings page. It is also push-to-talk.
Combined with Text-to-Speech, you can launch Agents with your voice and have the results read back to you!
On the Speech Settings page you can grant microphone access, test push-to-talk, and change the keyboard shortcut.
If you ever want to remove Ask Steve's microphone access, you can do it from: chrome://settings/content/microphone
BONUS: If you don't like the Ask Steve tab and buttons, you can permanently hide them on the Settings page and then just use the keyboard shortcut with your voice to launch your Agents.

This uses the browser's built in speech-recognition capability. Ask Steve takes the speech recognition result and sends it to an LLM with the names of all your Agents so it can figure out which one you're trying to launch.