Lead the model page with the two playable stereo results and expandable XML

README.md

@@ -30,6 +30,89 @@ model weights, the deterministic XML interleaver, the audio codec, voice
 presets, and inference code. It does not require an external transcript-generating
 language model, training caches, or a separate adapter/base checkpoint.
 
+## Featured examples and complete input XML
+
+These are the requested checkpoint-8565 recordings, normalized for listening.
+Both channels preserve their shared origin; neither lane is trimmed or shifted.
+
+### V01_S0192_I00000307-0000-student
+
+<audio controls src="https://huggingface.co/penguinfish1688/duplexdataengine/resolve/main/examples/V01_S0192_I00000307-0000-student.wav"></audio>
+
+[Download stereo audio](examples/V01_S0192_I00000307-0000-student.wav) · [Input XML](examples/V01_S0192_I00000307-0000.xml) · Voices: `voice_1` / `voice_2`
+
+<details>
+<summary>Show complete input XML</summary>
+
+```xml
+<duplex>
+<B>So, <e1/></B>
+<B start="e1">if you had a chance to <e2/></B>
+<B start="e2">send a message back to <e3/> the future, or the past, <e4/></B>
+<A start="e3"><sneeze/> <cough/></A>
+<B start="e4">or reverse, or whatever, however, what would you do? <e5/> Well, if I had a chance, I probably</B>
+<A start="e5">Well, if I had a chance, I'd probably. <e6/></A>
+<A start="e6"><e7/> Where can I do both, or I have to do one or the other? <e8/></A>
+<B start="e7">would do both. Probably,</B>
+<B start="e8">no, you probably will do both regardless. <e9/></B>
+<A start="e9">Well, most <e10/> likely, I'd send a message back to the past and be like, "Yo, <e11/> watch out! Don't go around that corner." <cough/> <e12/></A>
+<B start="e10">I most likely would</B>
+<B start="e11"><cough/></B>
+<B start="e12">who? <e13/> <laugh/></B>
+<A start="e13">I would send a message to myself. <e14/></A>
+<B start="e14">Oh, cool. <e15/></B>
+<A start="e15">I mean, who else would I contact in the past? <e16/></A>
+<B start="e16">Would you like yourself back then? <e17/></B>
+<A start="e17"><e18/> I don't listen. I have no fucking clue. <cough/> <e19/></A>
+<B start="e18"><cough/></B>
+<B start="e19">Okay. That far. Oh, great. <e20/> great! <e21/></B>
+<A start="e20">I don't have a clue.</A>
+<B start="e21"><gasp/> <sigh/> <sneeze/> Bye.</B>
+</duplex>
+```
+
+</details>
+
+### V01_S0220_I00000126-0000-student
+
+<audio controls src="https://huggingface.co/penguinfish1688/duplexdataengine/resolve/main/examples/V01_S0220_I00000126-0000-student.wav"></audio>
+
+[Download stereo audio](examples/V01_S0220_I00000126-0000-student.wav) · [Input XML](examples/V01_S0220_I00000126-0000.xml) · Voices: `voice_3` / `voice_4`
+
+<details>
+<summary>Show complete input XML</summary>
+
+```xml
+<duplex>
+<B>If we're to organize a fitness challenge, which <e1/></B>
+<B start="e1">I know you love, <e2/></B>
+<B start="e2">what would we include <e3/></B>
+<B start="e3">for activities? <e4/></B>
+<B start="e4">How would we make it engaging? <e5/></B>
+<B start="e5">What <e6/> do you think?</B>
+<A start="e6">The activities number one has got to be yoga, <e7/> probably.</A>
+<B start="e7">Yoga. I'm on <e8/> board. <e9/></B>
+<A start="e8">Stretching and</A>
+<A start="e9">you know, relaxing and feeling good. <e10/></A>
+<A start="e10">I I think weights are good too. I mean, just a little everything. <e11/></A>
+<A start="e11">I think one extreme to the other, maybe. <e12/></A>
+<B start="e12">Yeah, I mean, I'm definitely on board with yoga. I like mobility exercises. I <e13/></B>
+<A start="e13">Yeah, I think if you <e14/> do,</A>
+<B start="e14">don't know if I like the word extreme, but <e15/> you know.</B>
+<A start="e15">yeah. <e16/></A>
+<A start="e16">So the only thing missing would be so you have yoga, weights, and we need some kind of cardio. <e17/></A>
+<A start="e17"><e18/> Any suggestions? <e19/></A>
+<B start="e18">Yeah.</B>
+<B start="e19">Like speed walking, I guess. <e20/></B>
+<A start="e20">Maybe a treadmill. <e21/></A>
+<B start="e21">Yeah. Yeah, I like <e22/> the treadmill.</B>
+<A start="e22">Yeah, I think that would cover everything.</A>
+</duplex>
+```
+
+</details>
+
+
 ## Download and run
 
 Use Python 3.12 and a CUDA-capable NVIDIA GPU. Install matching PyTorch and
@@ -126,80 +209,6 @@ by example and channel, without inferring demographic characteristics.
 The released API accepts these preset IDs; arbitrary uploaded voice cloning is
 not part of this version.
 
-## Featured examples and complete input XML
-
-The following outputs are the requested checkpoint-8565 recordings. They use
-the same normalization as the inference API. Both channels have their original shared
-origin; neither lane is trimmed or shifted.
-
-### V01_S0192_I00000307-0000-student
-
-<audio controls src="https://huggingface.co/penguinfish1688/duplexdataengine/resolve/main/examples/V01_S0192_I00000307-0000-student.wav"></audio>
-
-[Download stereo audio](examples/V01_S0192_I00000307-0000-student.wav) · [Input XML](examples/V01_S0192_I00000307-0000.xml) · Voices: `voice_1` / `voice_2`
-
-```xml
-<duplex>
-<B>So, <e1/></B>
-<B start="e1">if you had a chance to <e2/></B>
-<B start="e2">send a message back to <e3/> the future, or the past, <e4/></B>
-<A start="e3"><sneeze/> <cough/></A>
-<B start="e4">or reverse, or whatever, however, what would you do? <e5/> Well, if I had a chance, I probably</B>
-<A start="e5">Well, if I had a chance, I'd probably. <e6/></A>
-<A start="e6"><e7/> Where can I do both, or I have to do one or the other? <e8/></A>
-<B start="e7">would do both. Probably,</B>
-<B start="e8">no, you probably will do both regardless. <e9/></B>
-<A start="e9">Well, most <e10/> likely, I'd send a message back to the past and be like, "Yo, <e11/> watch out! Don't go around that corner." <cough/> <e12/></A>
-<B start="e10">I most likely would</B>
-<B start="e11"><cough/></B>
-<B start="e12">who? <e13/> <laugh/></B>
-<A start="e13">I would send a message to myself. <e14/></A>
-<B start="e14">Oh, cool. <e15/></B>
-<A start="e15">I mean, who else would I contact in the past? <e16/></A>
-<B start="e16">Would you like yourself back then? <e17/></B>
-<A start="e17"><e18/> I don't listen. I have no fucking clue. <cough/> <e19/></A>
-<B start="e18"><cough/></B>
-<B start="e19">Okay. That far. Oh, great. <e20/> great! <e21/></B>
-<A start="e20">I don't have a clue.</A>
-<B start="e21"><gasp/> <sigh/> <sneeze/> Bye.</B>
-</duplex>
-```
-
-### V01_S0220_I00000126-0000-student
-
-<audio controls src="https://huggingface.co/penguinfish1688/duplexdataengine/resolve/main/examples/V01_S0220_I00000126-0000-student.wav"></audio>
-
-[Download stereo audio](examples/V01_S0220_I00000126-0000-student.wav) · [Input XML](examples/V01_S0220_I00000126-0000.xml) · Voices: `voice_3` / `voice_4`
-
-```xml
-<duplex>
-<B>If we're to organize a fitness challenge, which <e1/></B>
-<B start="e1">I know you love, <e2/></B>
-<B start="e2">what would we include <e3/></B>
-<B start="e3">for activities? <e4/></B>
-<B start="e4">How would we make it engaging? <e5/></B>
-<B start="e5">What <e6/> do you think?</B>
-<A start="e6">The activities number one has got to be yoga, <e7/> probably.</A>
-<B start="e7">Yoga. I'm on <e8/> board. <e9/></B>
-<A start="e8">Stretching and</A>
-<A start="e9">you know, relaxing and feeling good. <e10/></A>
-<A start="e10">I I think weights are good too. I mean, just a little everything. <e11/></A>
-<A start="e11">I think one extreme to the other, maybe. <e12/></A>
-<B start="e12">Yeah, I mean, I'm definitely on board with yoga. I like mobility exercises. I <e13/></B>
-<A start="e13">Yeah, I think if you <e14/> do,</A>
-<B start="e14">don't know if I like the word extreme, but <e15/> you know.</B>
-<A start="e15">yeah. <e16/></A>
-<A start="e16">So the only thing missing would be so you have yoga, weights, and we need some kind of cardio. <e17/></A>
-<A start="e17"><e18/> Any suggestions? <e19/></A>
-<B start="e18">Yeah.</B>
-<B start="e19">Like speed walking, I guess. <e20/></B>
-<A start="e20">Maybe a treadmill. <e21/></A>
-<B start="e21">Yeah. Yeah, I like <e22/> the treadmill.</B>
-<A start="e22">Yeah, I think that would cover everything.</A>
-</duplex>
-```
-
-
 ## Model and evaluation
 
 The shared model derives from Qwen3-TTS-12Hz-1.7B-CustomVoice. Stage III trained