ghostai1 commited on
Commit
c265174
Β·
verified Β·
1 Parent(s): 2c46e9f

Update index.html

Browse files
Files changed (1) hide show
  1. index.html +102 -69
index.html CHANGED
@@ -3,7 +3,7 @@
3
  <head>
4
  <meta charset="UTF-8" />
5
  <meta name="viewport" content="width=device-width, initial-scale=1" />
6
- <title>XTTVS-MED | Multilingual Voice-to-Voice Demo Suite</title>
7
 
8
  <!-- Bootstrap -->
9
  <link href="https://cdn.jsdelivr.net/npm/[email protected]/dist/css/bootstrap.min.css" rel="stylesheet"/>
@@ -11,136 +11,168 @@
11
  <!-- Mermaid -->
12
  <script src="https://cdn.jsdelivr.net/npm/mermaid@10/dist/mermaid.min.js"></script>
13
  <script>
14
- mermaid.initialize({
15
- startOnLoad:true,
16
- theme:'dark',
17
- flowchart:{fontSize:'16px'},
18
- sequence:{actorFontSize:'14px'}
19
- });
20
  </script>
21
 
22
  <style>
23
  :root{
24
- --bg:#070707; --panel:#111; --card:#181818;
25
- --txt:#e5e5e5; --accent:#00e5ff; --sub:#88f0ff;
26
  }
27
  *{box-sizing:border-box}
28
  body{margin:0;background:var(--bg);color:var(--txt);font-family:'Fira Code',monospace}
29
  a{color:var(--sub)} a:hover{text-decoration:underline}
30
- h1,h2,h3{color:var(--accent)}
31
- .container-lg{max-width:1160px;padding:2rem 1rem}
32
- .section{margin-bottom:3.2rem}
33
- .panel{background:var(--panel);padding:1.3rem;border-radius:.65rem}
34
  pre{
35
  background:var(--card);
36
- color:#c0f6ff;
37
- padding:1.2rem 1.7rem;
38
  border-radius:.55rem;
39
  font-size:.9rem;
40
  overflow-x:auto;
41
- margin:1.6rem 0;
42
  }
43
- code{background:#222;padding:.25rem .45rem;border-radius:.3rem}
44
  table{width:100%;border-collapse:collapse}
45
  th,td{padding:.8rem;border:1px solid #272727}
46
  th{background:var(--card);color:var(--accent)}
47
- tr:nth-child(even){background:#101010}
48
- .mermaid{background:var(--card);padding:1rem;border-radius:.55rem}
49
- footer{background:#0d0d0d;color:#888;text-align:center;padding:1.3rem 1rem;margin-top:4rem}
 
 
 
 
 
 
50
  </style>
51
  </head>
52
- <body>
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
53
 
54
- <div class="container-lg">
55
 
56
- <!-- Header -->
57
- <header class="text-center section">
58
- <h1>XTTVS-MED Demo Suite</h1>
59
- <p class="lead fw-semibold">
60
- Whisper β†’ Auto-Translate β†’ 4-bit XTTSv2 cloning&nbsp;Β·&nbsp;Sub-second across&nbsp;EN-AR-ES-FR
61
  </p>
62
  <p><strong>Chris Coleman</strong> β€” CEO / CTO Β· GhostAI Labs<br>
63
- <strong>Dr. Anthony Becker&nbsp;MD</strong> β€” Medical Advisor</p>
64
  </header>
65
 
66
- <!-- 1 -->
67
- <section class="section">
68
  <h2>1.&nbsp;Key Points</h2>
69
  <div class="panel">
70
  <ul class="mb-0">
71
- <li>⚑ End-to-end pipeline &lt; 1&nbsp;s on a 6&nbsp;GB gaming GPU.</li>
72
- <li>πŸ—£οΈ Same speaker timbre across four languages via LoRA adapters.</li>
73
- <li>πŸ“‰ 4-bit FloatBin quantization drops VRAM by 5-6Γ—.</li>
74
- <li>πŸ₯ Reduces critical ER communication delay by β‰ˆ 10-15 % survival uplift.</li>
75
  </ul>
76
  </div>
77
  </section>
78
 
79
  <!-- 2 Videos -->
80
- <section class="section">
81
- <h2>2.&nbsp;🎬 Demo Videos</h2>
82
 
83
- <h3 class="mt-4">2.1&nbsp;β€’&nbsp;Front-End Walk-Through (4&nbsp;Languages)</h3>
84
  <div class="ratio ratio-16x9 mb-3">
85
  <iframe src="https://streamable.com/e/bh1eff" allowfullscreen></iframe>
86
  </div>
87
 
88
- <h3 class="mt-5">2.2&nbsp;β€’&nbsp;Back-End Latency &amp; Hardware</h3>
89
  <div class="ratio ratio-16x9 mb-3">
90
  <iframe src="https://streamable.com/e/ryo2fv" allowfullscreen></iframe>
91
  </div>
92
  </section>
93
 
94
- <!-- 3 Architecture -->
95
- <section class="section">
96
- <h2>3.&nbsp;Pipeline (Time-over-Wavelength)</h2>
97
  <div class="mermaid">
98
  graph TD
99
- A["Input Audio"] --> B["Time-Wavelength<br>Decomposition"]
100
- B --> C["Semantic&nbsp;Vectors"]
101
- C --> D["FloatBin<br>Scheduler"]
102
- D --> E["XTTSv2<br>(INT4)"]
103
- E --> F["Cloned Output"]
104
  </div>
105
  </section>
106
 
107
- <!-- 4 Quant equation -->
108
- <section class="section">
109
  <h2>4.&nbsp;4-Bit FloatBin Quantization</h2>
110
- <pre>x<sub>q</sub> = round( (x βˆ’ x<sub>min</sub>) / (x<sub>max</sub> βˆ’ x<sub>min</sub>) Β· 15 )
111
  Β· (x<sub>max</sub> βˆ’ x<sub>min</sub>) / 15 + x<sub>min</sub></pre>
112
- <p class="small mb-0">Allows full model on GTX&nbsp;2060 6&nbsp;GB while keeping quality β‰₯ 4.4 MOS.</p>
113
  </section>
114
 
115
  <!-- 5 Hardware -->
116
- <section class="section">
117
  <h2>5.&nbsp;Latency vs. Hardware</h2>
118
  <table>
119
- <tr><th>System</th><th>Compute</th><th>VRAM</th><th>Latency&nbsp;250&nbsp;chars</th><th>Streams</th></tr>
120
- <tr><td>Pi&nbsp;5 + TPU</td><td>26 TFLOPS&nbsp;INT8</td><td>β€”</td><td>3.2 s</td><td>1-2</td></tr>
121
- <tr><td>RTX 2080</td><td>13 TFLOPS&nbsp;FP16</td><td>8 GB</td><td>1.2 s</td><td>3-4</td></tr>
122
- <tr><td>DGX A100</td><td>1 PFLOP</td><td>128 GB</td><td>0.4 s</td><td>20-30</td></tr>
123
- <tr><td>HF200 Cluster</td><td>2 PFLOPS</td><td>256 GB</td><td>0.2 s</td><td>40-50+</td></tr>
124
  </table>
 
125
  </section>
126
 
127
- <!-- 6 Use-case matrix -->
128
- <section class="section">
129
  <h2>6.&nbsp;Expanded Use-Case Matrix</h2>
130
  <table>
131
  <tr><th>Sector</th><th>Workflow</th><th>Impact</th></tr>
132
- <tr><td>ER&nbsp;Triage</td><td>Auto-voice vital alerts</td><td>&gt;&nbsp;10 % faster interventions</td></tr>
133
- <tr><td>Tele-ICU</td><td>Live captions + cloned voice</td><td>Lower staffing ratio</td></tr>
134
  <tr><td>Post-Op</td><td>Discharge voice reminders</td><td>↓ readmissions</td></tr>
135
  <tr><td>Pharmacy</td><td>Label read-outs</td><td>↑ adherence</td></tr>
136
- <tr><td>Mental-Health</td><td>Real-time interpreter</td><td>24/7 multilingual hotline</td></tr>
137
- <tr><td>Training</td><td>Procedures auto-narrated</td><td>Scalable global education</td></tr>
138
- <tr><td>Legal Consent</td><td>Forms voiced + display</td><td>Stronger audit trail</td></tr>
139
  </table>
140
  </section>
141
 
142
  <!-- 7 Fetch helper -->
143
- <section class="section">
144
  <h2>7.&nbsp;10-Line Fetch Helper</h2>
145
  <pre>
146
  // autoTranslateSpeak("I need pain medication.", "Spanish")
@@ -161,16 +193,17 @@ async function autoTranslateSpeak(text, target){
161
  </section>
162
 
163
  <!-- 8 Links -->
164
- <section class="section">
165
- <h2>8.&nbsp;Links</h2>
166
  <ul>
167
- <li>🌐 Live Space – <a href="https://huggingface.co/spaces/ghostai1/GHOSTVOICECBR" target="_blank">ghostvoicecbr on HF Spaces</a></li>
168
- <li>πŸ“Š Market brief – <a href="https://huggingface.co/spaces/ghostai1/GHOSTVOICECBR/blob/main/marketAImed.pdf" target="_blank">AImed market-fit</a></li>
 
169
  </ul>
170
  </section>
171
 
172
  <footer>
173
- &copy; 2025 GhostAI Labs Β· <a href="https://huggingface.co/spaces/ghostai1/GHOSTVOICECBR" target="_blank">Try the live app</a>
174
  </footer>
175
 
176
  </div>
 
3
  <head>
4
  <meta charset="UTF-8" />
5
  <meta name="viewport" content="width=device-width, initial-scale=1" />
6
+ <title>XTTVS-MED | Real-Time Multilingual Voice-to-Voice Suite</title>
7
 
8
  <!-- Bootstrap -->
9
  <link href="https://cdn.jsdelivr.net/npm/[email protected]/dist/css/bootstrap.min.css" rel="stylesheet"/>
 
11
  <!-- Mermaid -->
12
  <script src="https://cdn.jsdelivr.net/npm/mermaid@10/dist/mermaid.min.js"></script>
13
  <script>
14
+ mermaid.initialize({startOnLoad:true,theme:'dark',flowchart:{fontSize:'16px'},sequence:{actorFontSize:'14px'}});
 
 
 
 
 
15
  </script>
16
 
17
  <style>
18
  :root{
19
+ --bg:#070707; --panel:#101010; --card:#181818;
20
+ --txt:#e6e6e6; --accent:#00e5ff; --sub:#88f0ff;
21
  }
22
  *{box-sizing:border-box}
23
  body{margin:0;background:var(--bg);color:var(--txt);font-family:'Fira Code',monospace}
24
  a{color:var(--sub)} a:hover{text-decoration:underline}
25
+ h1,h2,h3,h4{color:var(--accent)}
26
+ .container-xl{max-width:1260px;padding:2rem 1rem}
27
+ .section{margin-bottom:3.3rem}
28
+ .panel{background:var(--panel);padding:1.3rem;border-radius:.7rem}
29
  pre{
30
  background:var(--card);
31
+ color:#c2f3ff;
32
+ padding:1.3rem 1.8rem;
33
  border-radius:.55rem;
34
  font-size:.9rem;
35
  overflow-x:auto;
36
+ margin:1.7rem 0;
37
  }
38
+ code{background:#242424;padding:.25rem .5rem;border-radius:.3rem}
39
  table{width:100%;border-collapse:collapse}
40
  th,td{padding:.8rem;border:1px solid #272727}
41
  th{background:var(--card);color:var(--accent)}
42
+ tr:nth-child(even){background:#0d0d0d}
43
+ .mermaid{background:var(--card);padding:1.1rem;border-radius:.6rem}
44
+ footer{background:#0d0d0d;color:#888;text-align:center;padding:1.4rem 1rem;margin-top:4rem;font-size:.9rem}
45
+ /* off-canvas menu button */
46
+ .navbar-toggler{border-color:var(--accent)}
47
+ .navbar-brand, .nav-link{color:var(--accent)!important}
48
+ .nav-link:hover{color:var(--sub)!important}
49
+ .offcanvas{background:var(--panel)}
50
+ .offcanvas a{color:var(--txt)!important}
51
  </style>
52
  </head>
53
+ <body data-bs-spy="scroll" data-bs-target="#navMain" data-bs-offset="100">
54
+
55
+ <!-- NAVBAR -->
56
+ <nav id="navMain" class="navbar navbar-dark navbar-expand-lg fixed-top" style="background:#000;">
57
+ <div class="container-fluid">
58
+ <a class="navbar-brand fw-bold" href="#">XTTVS-MED</a>
59
+ <button class="navbar-toggler" type="button" data-bs-toggle="offcanvas" data-bs-target="#offcanvasNav">
60
+ <span class="navbar-toggler-icon"></span>
61
+ </button>
62
+ </div>
63
+ </nav>
64
+
65
+ <!-- Off-canvas navigation -->
66
+ <div class="offcanvas offcanvas-end" tabindex="-1" id="offcanvasNav">
67
+ <div class="offcanvas-header">
68
+ <h5 class="offcanvas-title">Sections</h5>
69
+ <button class="btn-close btn-close-white" data-bs-dismiss="offcanvas"></button>
70
+ </div>
71
+ <div class="offcanvas-body">
72
+ <ul class="navbar-nav gap-2">
73
+ <li><a class="nav-link" href="#kp">Key Points</a></li>
74
+ <li><a class="nav-link" href="#videos">Demo Videos</a></li>
75
+ <li><a class="nav-link" href="#pipeline">Architecture</a></li>
76
+ <li><a class="nav-link" href="#quant">Quantization</a></li>
77
+ <li><a class="nav-link" href="#hardware">Hardware</a></li>
78
+ <li><a class="nav-link" href="#usecases">Use Cases</a></li>
79
+ <li><a class="nav-link" href="#fetch">Fetch Helper</a></li>
80
+ <li><a class="nav-link" href="#links">Links</a></li>
81
+ </ul>
82
+ </div>
83
+ </div>
84
 
85
+ <div class="container-xl">
86
 
87
+ <!-- Hero -->
88
+ <header class="text-center mb-5" style="padding-top:4.5rem">
89
+ <h1 class="mb-2">XTTVS-MED Demo Suite</h1>
90
+ <p class="lead fw-semibold mb-3">
91
+ Whisper&nbsp;β†’&nbsp;Translate&nbsp;β†’&nbsp;4-bit XTTSv2 | Sub-second EN-AR-ES-FR
92
  </p>
93
  <p><strong>Chris Coleman</strong> β€” CEO / CTO Β· GhostAI Labs<br>
94
+ <strong>Dr. Anthony Becker MD</strong> οΏ½οΏ½ Medical Advisor</p>
95
  </header>
96
 
97
+ <!-- 1 Key points -->
98
+ <section id="kp" class="section">
99
  <h2>1.&nbsp;Key Points</h2>
100
  <div class="panel">
101
  <ul class="mb-0">
102
+ <li>⚑ End-to-end &lt; 1 s on 6 GB consumer GPU.</li>
103
+ <li>πŸ—£οΈ Speaker timbre preserved across EN/AR/ES/FR via LoRA adapters.</li>
104
+ <li>🧩 FloatBin 4-bit quantization β†’ 5-6Γ— VRAM drop, same MOS.</li>
105
+ <li>πŸ₯ ER communication delay ↓ 10–15 % β†’ measurable survival uplift.</li>
106
  </ul>
107
  </div>
108
  </section>
109
 
110
  <!-- 2 Videos -->
111
+ <section id="videos" class="section">
112
+ <h2>2.&nbsp;🎬 Live Demo Videos</h2>
113
 
114
+ <h3 class="mt-4">2.1&nbsp;Front-End Walk-Through (4 Languages)</h3>
115
  <div class="ratio ratio-16x9 mb-3">
116
  <iframe src="https://streamable.com/e/bh1eff" allowfullscreen></iframe>
117
  </div>
118
 
119
+ <h3 class="mt-5">2.2&nbsp;Back-End Latency / Hardware</h3>
120
  <div class="ratio ratio-16x9 mb-3">
121
  <iframe src="https://streamable.com/e/ryo2fv" allowfullscreen></iframe>
122
  </div>
123
  </section>
124
 
125
+ <!-- 3 Pipeline -->
126
+ <section id="pipeline" class="section">
127
+ <h2>3.&nbsp;Architecture β€” Time-over-Wavelength Clone</h2>
128
  <div class="mermaid">
129
  graph TD
130
+ Mic["Input Audio"] --> TW["Time-Wavelength<br>Decompose"]
131
+ TW --> Sem["Semantic Vectors"]
132
+ Sem --> FB["FloatBin&nbsp;Scheduler"]
133
+ FB --> INT4["XTTS v2 (INT4)"]
134
+ INT4 --> Out["Cloned Output"]
135
  </div>
136
  </section>
137
 
138
+ <!-- 4 Quant -->
139
+ <section id="quant" class="section">
140
  <h2>4.&nbsp;4-Bit FloatBin Quantization</h2>
141
+ <pre>x<sub>q</sub> = round( (x βˆ’ x<sub>min</sub>) / (x<sub>max</sub> βˆ’ x<sub>min</sub>) Β· 15 )
142
  Β· (x<sub>max</sub> βˆ’ x<sub>min</sub>) / 15 + x<sub>min</sub></pre>
143
+ <p class="small mb-0">Works on GTX 2060 6 GB with MOS β‰₯ 4.4 and 0.8 s latency.</p>
144
  </section>
145
 
146
  <!-- 5 Hardware -->
147
+ <section id="hardware" class="section">
148
  <h2>5.&nbsp;Latency vs. Hardware</h2>
149
  <table>
150
+ <tr><th>System</th><th>Compute</th><th>VRAM</th><th>Latency* (250 chars)</th><th>Streams</th></tr>
151
+ <tr><td>Pi 5 + Edge TPU</td><td>26 TFLOPS INT8</td><td>β€”</td><td>3.2 s</td><td>1–2</td></tr>
152
+ <tr><td>RTX 2080</td><td>13 TFLOPS FP16</td><td>8 GB</td><td>1.2 s</td><td>3–4</td></tr>
153
+ <tr><td>DGX A100</td><td>1 PFLOP</td><td>128 GB</td><td>0.4 s</td><td>20–30</td></tr>
154
+ <tr><td>HF200 Cluster</td><td>2 PFLOPS</td><td>256 GB</td><td>0.2 s</td><td>40–50+</td></tr>
155
  </table>
156
+ <p class="small">*End-to-end: ASR β†’ Translate β†’ Speech synthesis</p>
157
  </section>
158
 
159
+ <!-- 6 Use cases -->
160
+ <section id="usecases" class="section">
161
  <h2>6.&nbsp;Expanded Use-Case Matrix</h2>
162
  <table>
163
  <tr><th>Sector</th><th>Workflow</th><th>Impact</th></tr>
164
+ <tr><td>ER Triage</td><td>Vitals voiced in patient language</td><td>&gt; 10 % faster intervention</td></tr>
165
+ <tr><td>Tele-ICU</td><td>Live caption + cloned voice</td><td>Lower staff ratio</td></tr>
166
  <tr><td>Post-Op</td><td>Discharge voice reminders</td><td>↓ readmissions</td></tr>
167
  <tr><td>Pharmacy</td><td>Label read-outs</td><td>↑ adherence</td></tr>
168
+ <tr><td>Mental-Health</td><td>Interpreter in crisis hotlines</td><td>24/7 multilingual support</td></tr>
169
+ <tr><td>Training</td><td>Procedures auto-narrated</td><td>Global education scale</td></tr>
170
+ <tr><td>Legal Consent</td><td>Forms voiced &amp; displayed</td><td>Stronger audit trail</td></tr>
171
  </table>
172
  </section>
173
 
174
  <!-- 7 Fetch helper -->
175
+ <section id="fetch" class="section">
176
  <h2>7.&nbsp;10-Line Fetch Helper</h2>
177
  <pre>
178
  // autoTranslateSpeak("I need pain medication.", "Spanish")
 
193
  </section>
194
 
195
  <!-- 8 Links -->
196
+ <section id="links" class="section">
197
+ <h2>8.&nbsp;Try It / Learn More</h2>
198
  <ul>
199
+ <li>🌐 Live Space β€” <a href="https://huggingface.co/spaces/ghostai1/GHOSTVOICECBR" target="_blank">ghostvoicecbr (Hugging Face)</a></li>
200
+ <li>πŸ“„ Architecture deep-dive β€” <a href="https://huggingface.co/spaces/ghostai1/GHOSTVOICECBR/blob/main/ghostcbrwpaper2025.pdf" target="_blank">White-Paper 2025</a></li>
201
+ <li>πŸ“ˆ Market rationale β€” <a href="https://huggingface.co/spaces/ghostai1/GHOSTVOICECBR/blob/main/marketAImed.pdf" target="_blank">AI-Med brief</a></li>
202
  </ul>
203
  </section>
204
 
205
  <footer>
206
+ Β© 2025 GhostAI Labs Β· <a href="https://huggingface.co/spaces/ghostai1/GHOSTVOICECBR" target="_blank">Launch the live app</a>
207
  </footer>
208
 
209
  </div>