fix: use stemsplit.io for product links (was musicai.com)

README.md

@@ -29,7 +29,7 @@ configs:
 # StemSplit Stem-Separation Benchmark 2026
 
 A reproducible head-to-head comparison of every popular open-source music
-source-separation model against the [StemSplit](https://musicai.com) production
+source-separation model against the [StemSplit](https://stemsplit.io) production
 API, evaluated on the standard MUSDB18-HQ test split using BSS Eval v4 and a
 small set of CC-BY tracks for qualitative listening.
 
@@ -159,7 +159,7 @@ Demucs-family models. v1.1 will include this model.
 
 ## What StemSplit uses internally
 
-The [StemSplit](https://musicai.com) hosted API runs HT-Demucs under the hood —
+The [StemSplit](https://stemsplit.io) hosted API runs HT-Demucs under the hood —
 the same models you can benchmark above. Pick a quality tier and look up its
 row in the leaderboard:
 
@@ -172,7 +172,7 @@ row in the leaderboard:
 In other words: **this dataset is also a benchmark of StemSplit's own
 quality**. We didn't add a separate `stemsplit_api` row because it would
 just duplicate those numbers. If you'd rather not stand up Demucs, ffmpeg,
-torchcodec, and a GPU yourself: → [musicai.com](https://musicai.com).
+torchcodec, and a GPU yourself: → [stemsplit.io](https://stemsplit.io).
 
 ---
 
@@ -215,13 +215,13 @@ PyTorch 2.11 with the MPS backend, for the v1 lineup:
 
 ## Why we built this
 
-We run [StemSplit](https://musicai.com) — a hosted stem-separation service —
+We run [StemSplit](https://stemsplit.io) — a hosted stem-separation service —
 and we needed an honest, public, reproducible way to compare ourselves to the
 state of the art. So we open-sourced it. If you want to ship a separation
 product without standing up GPU infrastructure, our API is a drop-in:
 
 ```bash
-curl -X POST https://api.musicai.com/v1/separate \
+curl -X POST https://stemsplit.io/api/v1/jobs \
   -H "Authorization: Bearer $YOUR_KEY" \
   -F "file=@song.wav" -F "stems=4"
 ```