Add kitten

model.py

@@ -119,6 +119,51 @@ def _get_vits_ljs(repo_id: str, speed: float) -> sherpa_onnx.OfflineTts:
     return tts
 
 
+@lru_cache(maxsize=10)
+def _get_kitten(repo_id: str, speed: float) -> sherpa_onnx.OfflineTts:
+    data_dir = "/tmp/espeak-ng-data"
+    repo_id = repo_id.split("|")[0]
+    assert repo_id in ("csukuangfj/kitten-nano-en-v0_1-fp16",), repo_id
+
+    model = get_file(
+        repo_id=repo_id,
+        filename="model.fp16.onnx",
+        subfolder=".",
+    )
+
+    tokens = get_file(
+        repo_id=repo_id,
+        filename="tokens.txt",
+        subfolder=".",
+    )
+
+    voices = get_file(
+        repo_id=repo_id,
+        filename="voices.bin",
+        subfolder=".",
+    )
+
+    tts_config = sherpa_onnx.OfflineTtsConfig(
+        model=sherpa_onnx.OfflineTtsModelConfig(
+            kitten=sherpa_onnx.OfflineTtsKittenModelConfig(
+                model=model,
+                voices=voices,
+                tokens=tokens,
+                data_dir=data_dir,
+                length_scale=1.0 / speed,
+            ),
+            provider="cpu",
+            debug=True,
+            num_threads=2,
+        ),
+        max_num_sentences=1,
+    )
+
+    tts = sherpa_onnx.OfflineTts(tts_config)
+
+    return tts
+
+
 @lru_cache(maxsize=10)
 def _get_kokoro(repo_id: str, speed: float) -> sherpa_onnx.OfflineTts:
     data_dir = "/tmp/espeak-ng-data"
@@ -695,6 +740,7 @@ chinese_models = {
 
 english_models = {
     "csukuangfj/kokoro-en-v0_19|11 speakers": _get_kokoro,
+    "csukuangfj/kitten-nano-en-v0_1-fp16|8 speakers": _get_kitten,
     "csukuangfj/vits-piper-en_US-glados-high|1 speaker": _get_vits_piper,
     "csukuangfj/vits-piper-en_US-glados|1 speaker": _get_vits_piper,
     "csukuangfj/vits-piper-en_GB-southern_english_male-medium|8 speakers": _get_vits_piper,