update whisper

Author: Abandon-ht

projects/llm_framework/main_whisper/Kconfig

@@ -0,0 +1,49 @@
+import os
+
+Import('env')
+with open(env['PROJECT_TOOL_S']) as f:
+    exec(f.read())
+
+# env.Append(CXXFLAGS=['-O3', '-fopenmp', '-std=c++17'])
+SRCS = append_srcs_dir(ADir('src'))
+INCLUDE = [ADir('include'), ADir('.')]
+PRIVATE_INCLUDE = []
+REQUIREMENTS = ['pthread', 'utilities', 'ax_msp', 'eventpp', 'StackFlow', 'single_header_libs']
+STATIC_LIB = []
+DYNAMIC_LIB = []
+DEFINITIONS = []
+DEFINITIONS_PRIVATE = []
+LDFLAGS = []
+LINK_SEARCH_PATH = []
+STATIC_FILES = []
+
+DEFINITIONS += ['-O3', '-fopenmp', '-std=c++17']
+LDFLAGS+=['-Wl,-rpath=/opt/m5stack/lib', '-Wl,-rpath=/usr/local/m5stack/lib', '-Wl,-rpath=/usr/local/m5stack/lib/gcc-10.3', '-Wl,-rpath=/opt/lib', '-Wl,-rpath=/opt/usr/lib', '-Wl,-rpath=./']
+LINK_SEARCH_PATH += [ADir('../static_lib')]
+REQUIREMENTS += ['ax_engine', 'ax_interpreter', 'ax_sys']
+REQUIREMENTS += ['onnxruntime', 'samplerate']
+
+INCLUDE += [ADir('../include')]
+INCLUDE += [ADir('src/runner'), ADir('../include/onnxruntime/core/session')]
+INCLUDE += [ADir('src/runner/opencc/include/opencc')]
+
+LINK_SEARCH_PATH += [ADir('src/runner/opencc/lib')]
+# LDFLAGS += ['-l:libcargs.a', '-l:libonnxruntime.a']
+LDFLAGS += ['-l:libopencc.a', '-l:libmarisa.a']
+
+STATIC_FILES += Glob('mode_*.json')
+
+env['COMPONENTS'].append({'target':'llm_whisper',
+                          'SRCS':SRCS,
+                          'INCLUDE':INCLUDE,
+                          'PRIVATE_INCLUDE':PRIVATE_INCLUDE,
+                          'REQUIREMENTS':REQUIREMENTS,
+                          'STATIC_LIB':STATIC_LIB,
+                          'DYNAMIC_LIB':DYNAMIC_LIB,
+                          'DEFINITIONS':DEFINITIONS,
+                          'DEFINITIONS_PRIVATE':DEFINITIONS_PRIVATE,
+                          'LDFLAGS':LDFLAGS,
+                          'LINK_SEARCH_PATH':LINK_SEARCH_PATH,
+                          'STATIC_FILES':STATIC_FILES,
+                          'REGISTER':'project'
+                          })

projects/llm_framework/main_whisper/SConstruct

@@ -0,0 +1,40 @@
+{
+    "mode": "whisper-tiny",
+    "type": "asr",
+    "capabilities": [
+        "Automatic_Speech_Recognition",
+        "English",
+        "Chinese"
+    ],
+    "input_type": [
+        "sys.pcm"
+    ],
+    "output_type": [
+        "asr.utf-8"
+    ],
+    "mode_param": {
+        "model_type": "tiny",
+        "language": "zh",
+        "encoder": "tiny-encoder.axmodel",
+        "decoder_main": "tiny-decoder-main.axmodel",
+        "decoder_loop": "tiny-decoder-loop.axmodel",
+        "position_embedding": "tiny-positional_embedding.bin",
+        "tokens": "tiny-tokens.txt",
+        "t2s": "t2s.json",
+        "whisper_sample_rate": 16000,
+        "whisper_n_fft": 400,
+        "awake_delay": 1000,
+        "whisper_hop_length": 160,
+        "whisper_chunk_size": 30,
+        "whisper_n_mels": 80,
+        "whisper_sot": 50258,
+        "whisper_eot": 50257,
+        "whisper_blank": 220,
+        "whisper_no_timestamps": 50363,
+        "whisper_no_speech": 50362,
+        "whisper_translate": 50358,
+        "whisper_transcribe": 50359,
+        "whisper_vocab_size": 51865,
+        "whisper_n_text_ctx": 448
+    }
+}