You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit 1a327fe
Browse filesBrowse the repository at this point in the historyBrowse files
feat: homepage features voice chat with your agents; better agent voices (0.13.1, qc 0.9.1) (#331)
Anthony: "update the homepage to feature voice chat with your agents...invite your
agents to a group and name them", and on the call audio: "it sounds like someone
using a voicebox".
- Homepage: a featured "Voice chat with your AI agents" section under the hero
(invite your agents, name them and add them to a group, tap the mic) with a
sample call, a feature card and a JSON-LD featureList entry. Also fixes the
phone layout: the nowrap install command made the hero 556 px wide on a 390 px
screen (grid items now min-width: 0).
- qc agent call --tts <url>: speak with your own voice server (POST { text, voice }
-> WAV), falling back to qrypt.chat's Aura voice if it fails. dev2 runs Kokoro.
- Agent audio is published at up to 64 kbps Opus with DTX off and RED on, so
synthetic speech is not clipped into a robotic sound.
Measured before changing anything: the call path itself is clean (no dropouts or
zero runs inside speech, pitch intact at 250 Hz), so the fix is the voice and
the bitrate.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Copy file name to clipboardExpand all lines: packages/qryptchat/package.json
+1-1Lines changed: 1 addition & 1 deletion
Original file line number
Diff line number
Diff line change
@@ -1,6 +1,6 @@
1
1
{
2
2
"name": "@profullstack/qryptchat",
3
-
"version": "0.9.0",
3
+
"version": "0.9.1",
4
4
"description": "qc: qrypt.chat in your terminal. A full-screen end-to-end encrypted chat client (ML-KEM-1024), plus a scriptable CLI and an MCP server.",
<h2id="agents-voice-title">Voice chat with your AI agents</h2>
60
+
<pclassName="agents-voice-lede">
61
+
Invite your agents to a group, give them names, and just talk. They join the call, listen,
62
+
take turns like people, and answer out loud in their own voices. Ask for real work and they
63
+
get on it, then tell you how it went.
64
+
</p>
65
+
<olclassName="agents-voice-steps">
66
+
<li><strong>Invite your agents.</strong> Settings → AI agents sends an invite by link, email or text; an agent joins with <code>qc agent join</code> and makes its own keys.</li>
67
+
<li><strong>Name them and add them to a group.</strong> One agent for code, one for your inbox, whatever you run: they each know their job.</li>
68
+
<li><strong>Tap the mic.</strong> A voice call starts, your agents join, and you talk. Everything they say is also written into the chat.</li>
69
+
</ol>
70
+
<Linkhref="/auth"className="btn btn-primary agents-voice-cta">Start talking to your agents</Link>
@@ -64,6 +92,7 @@ export default function HomeContent() {
64
92
{num: '03',title: '🔐 Encrypted Voice & Video Calls',desc: 'Crystal-clear 1:1 and group calls protected with ML-KEM-1024 post-quantum encryption.'},
65
93
{num: '04',title: 'Real-time Messaging',desc: 'Instant message delivery with typing indicators and read receipts.'},
66
94
{num: '05',title: 'Multi-language Support',desc: 'Available in multiple languages with full internationalization support.'},
95
+
{num: '06',title: '🤖 Voice Chat With Your AI Agents',desc: 'Invite your agents to a group, name them, and talk: they join an encrypted call and answer in their own voices.'},
67
96
].map(({ num, title, desc })=>(
68
97
<divkey={num}className="feature-item">
69
98
<divclassName="feature-number">{num}</div>
@@ -100,6 +129,25 @@ export default function HomeContent() {
100
129
.steps li { color: var(--color-text-secondary); line-height: 1.7; font-size: 1.0625rem; }
101
130
.steps li strong { color: var(--color-text-primary); }
0 commit comments