Document automatic stress and cross-language duration guidance

README.md

@@ -18,6 +18,13 @@ streamed audio output.
 This release uses the clean English/Russian 25-second teacher and its matching
 eight-step CFG-3 distilled student.
 
+> **Important — Russian stress is automatic.** Text inside `<ru>…</ru>` receives
+> stress markers automatically by default. Explicit `+` markers always win.
+>
+> **Important — cross-language prompts.** When an English reference voice is
+> speaking Russian, experiment with `duration_scale` below `1` (for example
+> `0.8`). It is usually a better starting point than the default `1`.
+
 ## Installation
 
 ```bash
@@ -107,6 +114,17 @@ words and applies deterministic `ё` replacements, while unknown words and
 ambiguous homographs are left unchanged. Set `russian_stress=False` to disable
 automatic Russian stress processing entirely.
 
+When using an English voice such as `eng_f3` for Russian text, start by trying
+`duration_scale=0.8` and adjust by ear:
+
+```python
+waveform = tts.generate_speech(
+    "<ru>Это русский текст английским голосом.</ru>",
+    voice="eng_f3",
+    duration_scale=0.8,
+)
+```
+
 ## Stream audio
 
 ```python

config.json

@@ -1,11 +1,13 @@
 {
   "architectures": ["TeraTTSModel"],
+  "automatic_russian_stress": true,
   "auto_map": {
     "AutoConfig": "configuration_teratts.TeraTTSConfig",
     "AutoModel": "modeling_teratts.TeraTTSModel"
   },
   "default_diffusion_model": "distilled",
   "default_voice": "ru_f1",
+  "cross_language_prompt_note": "For an English reference voice speaking Russian, try duration_scale below 1.0.",
   "model_type": "teratts_onnx",
   "sample_rate": 44100,
   "voices": ["eng_f3", "eng_f4_whisper", "eng_f5", "eng_m2_whisper", "eng_m3", "eng_m4", "ru_f1", "ru_f2", "ru_m1", "ru_m5"]

configuration_teratts.py

@@ -12,6 +12,10 @@ class TeraTTSConfig(PretrainedConfig):
         default_diffusion_model: str = "distilled",
         default_voice: str = "ru_f1",
         voices: list[str] | None = None,
+        automatic_russian_stress: bool = True,
+        cross_language_prompt_note: str = (
+            "For an English reference voice speaking Russian, try duration_scale below 1.0."
+        ),
         **kwargs,
     ) -> None:
         super().__init__(**kwargs)
@@ -19,3 +23,5 @@ class TeraTTSConfig(PretrainedConfig):
         self.default_diffusion_model = default_diffusion_model
         self.default_voice = default_voice
         self.voices = voices or []
+        self.automatic_russian_stress = automatic_russian_stress
+        self.cross_language_prompt_note = cross_language_prompt_note