From 52e951b32551f37299afc596c9cd3e8e20354054 Mon Sep 17 00:00:00 2001
From: LittleMouse <niuchenyang@m5stack.com>
Date: Fri, 26 Dec 2025 16:56:23 +0800
Subject: [PATCH] [update] supported sensevoice asr

---
 backend/asr_client_backend.py | 13 +++++++------
 backend/tts_client_backend.py |  2 +-
 services/model_list.py        | 12 ++++++------
 3 files changed, 14 insertions(+), 13 deletions(-)

diff --git a/backend/asr_client_backend.py b/backend/asr_client_backend.py
index d64b30c..d9998ba 100644
--- a/backend/asr_client_backend.py
+++ b/backend/asr_client_backend.py
@@ -58,16 +58,17 @@ class ASRClientBackend(BaseModelBackend):
 
             loop = asyncio.get_event_loop()
             await loop.run_in_executor(
-                None,
-                client.setup,
-                "whisper.setup",
-                {
+                self._inference_executor,
+                lambda: client.setup(
+                    self.config["object"],
+                    {
                     "model": self.config["model_name"],
                     "response_format": "asr.utf-8",
-                    "input": "whisper.wav.stream.base64",
+                    "input": "whisper.asr.wav.stream.base64",
                     "language": language,
                     "enoutput": True
-                }
+                    }
+                )
             )
             return client
         except asyncio.TimeoutError:
diff --git a/backend/tts_client_backend.py b/backend/tts_client_backend.py
index 8d74a29..e840e6b 100644
--- a/backend/tts_client_backend.py
+++ b/backend/tts_client_backend.py
@@ -30,7 +30,7 @@ class TtsClientBackend(BaseModelBackend):
             host=self.config["host"],
             port=self.config["port"]
         )
-        self.sample_rate =  model_config.get("sample_rate", 16000)
+        self.sample_rate =  model_config.get("audio_rate", 16000)
         self.channels = 1
         self.tokenizer = tiktoken.get_encoding("cl100k_base")
 
diff --git a/services/model_list.py b/services/model_list.py
index c81bfce..7c4dbd2 100644
--- a/services/model_list.py
+++ b/services/model_list.py
@@ -99,22 +99,24 @@ class GetModelList:
                     elif model_type == 'tts':
                         mode_param = model_data.get("mode_param", {})
                         precompute_len = None
-                        sample_rate = None
+                        audio_rate = None
                         cmm_size = None
                         if isinstance(mode_param, dict):
                             precompute_len = mode_param.get("precompute_len")
                             cmm_size = mode_param.get("cmm_size")
-                            sample_rate = mode_param.get("sample_rate")
+                            audio_rate = mode_param.get("audio_rate")
 
                         if 'melotts' in mode.lower():
                             obj = 'melotts.setup'
                             new_entry['memory_required'] = 59764
-                            new_entry['sample_rate'] = 16000
+                            new_entry['audio_rate'] = 16000
                             new_entry['max_context_length'] = 32768
+                            if audio_rate is not None:
+                                new_entry['audio_rate'] = audio_rate
                         elif 'cosyvoice' in mode.lower():
                             obj = 'cosy_voice.setup'
                             new_entry['memory_required'] = 1185772
-                            new_entry['sample_rate'] = 48000
+                            new_entry['audio_rate'] = 48000
                         else:
                             obj = 'tts.setup'
 
@@ -122,8 +124,6 @@ class GetModelList:
                             new_entry['memory_required'] = cmm_size
                         if precompute_len is not None:
                             new_entry['max_context_length'] = precompute_len
-                        if sample_rate is not None:
-                            new_entry['max_context_length'] = sample_rate
 
                         new_entry.update({
                             "response_format": "wav.base64",