Merge pull request #3260 from xinnan-tech/test-funasr

FunASR识别指定语言

main/manager-api/src/main/resources/db/changelog/202607011405.sql

@@ -0,0 +1,19 @@
+-- Add language configuration for local FunASR ASR.
+UPDATE `ai_model_provider`
+SET `fields` = '[{"key":"model_dir","label":"模型目录","type":"string"},{"key":"output_dir","label":"输出目录","type":"string"},{"key":"language","label":"识别语言","type":"string","default":"auto"}]'
+WHERE `id` = 'SYSTEM_ASR_FunASR';
+
+UPDATE `ai_model_config`
+SET `config_json` = JSON_SET(`config_json`, '$.language', 'auto')
+WHERE `id` = 'ASR_FunASR'
+AND JSON_EXTRACT(`config_json`, '$.language') IS NULL;
+
+-- Update the FunASR local model configuration description to mention the language option.
+UPDATE `ai_model_config`
+SET `remark` = 'FunASR本地模型配置说明：
+1. 需要下载模型文件到xiaozhi-server/models/SenseVoiceSmall目录
+2. 支持中日韩粤语音识别
+3. 本地推理，无需网络连接
+4. 待识别文件保存在tmp/目录
+5. “识别语言”字段控制识别语种：auto = 自动检测；如需限定只识别中文可设为 zh（en=英语、ja=日语、ko=韩语、yue=粤语）。'
+WHERE `id` = 'ASR_FunASR';

main/manager-api/src/main/resources/db/changelog/db.changelog-master.yaml

@@ -690,3 +690,10 @@ databaseChangeLog:
         - sqlFile:
             encoding: utf8
             path: classpath:db/changelog/202606261131.sql
+  - changeSet:
+      id: 202607011405
+      author: LauraGPT
+      changes:
+        - sqlFile:
+            encoding: utf8
+            path: classpath:db/changelog/202607011405.sql

main/xiaozhi-server/config.yaml

@@ -360,6 +360,9 @@ ASR:
     type: fun_local
     model_dir: models/SenseVoiceSmall
     output_dir: tmp/
+    # 识别语种：auto 自动检测；如需限制只识别中文可设为 zh，避免中文短句误判为韩文等。
+    # SenseVoice 支持 zh、en、ja、ko、yue 等语种标记。
+    language: auto
   FunASRServer:
     # 独立部署FunASR，使用FunASR的API服务，只需要五句话
     # 第一句：mkdir -p ./funasr-runtime-resources/models

main/xiaozhi-server/core/providers/asr/fun_local.py

@@ -50,6 +50,7 @@ class ASRProvider(ASRProviderBase):
         self.interface_type = InterfaceType.LOCAL
         self.model_dir = config.get("model_dir")
         self.output_dir = config.get("output_dir")  # 修正配置键名
+        self.language = config.get("language", "auto")
         self.delete_audio_file = delete_audio_file
 
         # 确保输出目录存在
@@ -80,7 +81,7 @@ class ASRProvider(ASRProviderBase):
                     self.model.generate,
                     input=artifacts.pcm_bytes,
                     cache={},
-                    language="auto",
+                    language=self.language,
                     use_itn=True,
                     batch_size_s=60,
                 )