Commit
·
387e4e6
1
Parent(s):
320f57c
add more models
Browse files
model.py
CHANGED
@@ -170,6 +170,10 @@ def get_pretrained_model(
|
|
170 |
return multi_lingual_models[repo_id](
|
171 |
repo_id, decoding_method=decoding_method, num_active_paths=num_active_paths
|
172 |
)
|
|
|
|
|
|
|
|
|
173 |
elif repo_id in chinese_models:
|
174 |
return chinese_models[repo_id](
|
175 |
repo_id, decoding_method=decoding_method, num_active_paths=num_active_paths
|
@@ -1594,12 +1598,14 @@ def _get_streaming_zipformer_pre_trained_model(
|
|
1594 |
"csukuangfj/sherpa-onnx-streaming-zipformer-en-kroko-2025-08-06",
|
1595 |
"csukuangfj/sherpa-onnx-streaming-zipformer-es-kroko-2025-08-06",
|
1596 |
"csukuangfj/sherpa-onnx-streaming-zipformer-fr-kroko-2025-08-06",
|
|
|
1597 |
], repo_id
|
1598 |
|
1599 |
if repo_id in [
|
1600 |
"csukuangfj/sherpa-onnx-streaming-zipformer-en-kroko-2025-08-06",
|
1601 |
"csukuangfj/sherpa-onnx-streaming-zipformer-es-kroko-2025-08-06",
|
1602 |
"csukuangfj/sherpa-onnx-streaming-zipformer-fr-kroko-2025-08-06",
|
|
|
1603 |
]:
|
1604 |
encoder_model = _get_nn_model_filename(
|
1605 |
repo_id=repo_id,
|
@@ -2089,6 +2095,10 @@ english_models = {
|
|
2089 |
"csukuangfj/wenet-english-model": _get_wenet_model,
|
2090 |
}
|
2091 |
|
|
|
|
|
|
|
|
|
2092 |
multi_lingual_models = {
|
2093 |
"csukuangfj/sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02": _get_dolphin_ctc_models,
|
2094 |
"csukuangfj/sherpa-onnx-dolphin-small-ctc-multi-lang-int8-2025-04-02": _get_dolphin_ctc_models,
|
@@ -2115,6 +2125,8 @@ arabic_models = {
|
|
2115 |
}
|
2116 |
|
2117 |
german_models = {
|
|
|
|
|
2118 |
"csukuangfj/sherpa-onnx-nemo-transducer-stt_de_fastconformer_hybrid_large_pc": _get_sherpa_onnx_nemo_transducer_models,
|
2119 |
"csukuangfj/sherpa-onnx-nemo-transducer-stt_de_fastconformer_hybrid_large_pc-int8": _get_sherpa_onnx_nemo_transducer_models_int8,
|
2120 |
"csukuangfj/sherpa-onnx-nemo-stt_de_fastconformer_hybrid_large_pc": _get_sherpa_onnx_nemo_ctc_models,
|
@@ -2123,11 +2135,13 @@ german_models = {
|
|
2123 |
}
|
2124 |
|
2125 |
french_models = {
|
|
|
2126 |
"shaojieli/sherpa-onnx-streaming-zipformer-fr-2023-04-14": _get_french_pre_trained_model,
|
2127 |
"csukuangfj/sherpa-onnx-streaming-zipformer-fr-kroko-2025-08-06": _get_streaming_zipformer_pre_trained_model,
|
2128 |
}
|
2129 |
|
2130 |
spanish_models = {
|
|
|
2131 |
"csukuangfj/sherpa-onnx-streaming-zipformer-es-kroko-2025-08-06": _get_streaming_zipformer_pre_trained_model,
|
2132 |
}
|
2133 |
|
@@ -2138,6 +2152,7 @@ japanese_models = {
|
|
2138 |
}
|
2139 |
|
2140 |
russian_models = {
|
|
|
2141 |
"csukuangfj/sherpa-onnx-nemo-transducer-giga-am-v2-russian-2025-04-19": _get_russian_pre_trained_model,
|
2142 |
"csukuangfj/sherpa-onnx-nemo-ctc-giga-am-v2-russian-2025-04-19": _get_russian_pre_trained_model_ctc,
|
2143 |
"csukuangfj/sherpa-onnx-nemo-transducer-giga-am-russian-2024-10-24": _get_russian_pre_trained_model,
|
@@ -2174,6 +2189,7 @@ vietnamese_models = {
|
|
2174 |
}
|
2175 |
|
2176 |
portuguese_brazlian_models = {
|
|
|
2177 |
"csukuangfj/sherpa-onnx-nemo-stt_pt_fastconformer_hybrid_large_pc": _get_sherpa_onnx_nemo_ctc_models,
|
2178 |
"csukuangfj/sherpa-onnx-nemo-stt_pt_fastconformer_hybrid_large_pc-int8": _get_sherpa_onnx_nemo_ctc_models,
|
2179 |
"csukuangfj/sherpa-onnx-nemo-transducer-stt_pt_fastconformer_hybrid_large_pc": _get_sherpa_onnx_nemo_transducer_models,
|
@@ -2182,6 +2198,7 @@ portuguese_brazlian_models = {
|
|
2182 |
|
2183 |
|
2184 |
all_models = {
|
|
|
2185 |
**multi_lingual_models,
|
2186 |
**chinese_models,
|
2187 |
**english_models,
|
@@ -2203,6 +2220,7 @@ all_models = {
|
|
2203 |
}
|
2204 |
|
2205 |
language_to_models = {
|
|
|
2206 |
"Multi-lingual (east aisa)": list(multi_lingual_models.keys()),
|
2207 |
"超多种中文方言": list(chinese_dialect_models.keys()),
|
2208 |
"Chinese": list(chinese_models.keys()),
|
|
|
170 |
return multi_lingual_models[repo_id](
|
171 |
repo_id, decoding_method=decoding_method, num_active_paths=num_active_paths
|
172 |
)
|
173 |
+
elif repo_id in twenty_five_languages_models:
|
174 |
+
return twenty_five_languages_models[repo_id](
|
175 |
+
repo_id, decoding_method=decoding_method, num_active_paths=num_active_paths
|
176 |
+
)
|
177 |
elif repo_id in chinese_models:
|
178 |
return chinese_models[repo_id](
|
179 |
repo_id, decoding_method=decoding_method, num_active_paths=num_active_paths
|
|
|
1598 |
"csukuangfj/sherpa-onnx-streaming-zipformer-en-kroko-2025-08-06",
|
1599 |
"csukuangfj/sherpa-onnx-streaming-zipformer-es-kroko-2025-08-06",
|
1600 |
"csukuangfj/sherpa-onnx-streaming-zipformer-fr-kroko-2025-08-06",
|
1601 |
+
"csukuangfj/sherpa-onnx-streaming-zipformer-de-kroko-2025-08-06",
|
1602 |
], repo_id
|
1603 |
|
1604 |
if repo_id in [
|
1605 |
"csukuangfj/sherpa-onnx-streaming-zipformer-en-kroko-2025-08-06",
|
1606 |
"csukuangfj/sherpa-onnx-streaming-zipformer-es-kroko-2025-08-06",
|
1607 |
"csukuangfj/sherpa-onnx-streaming-zipformer-fr-kroko-2025-08-06",
|
1608 |
+
"csukuangfj/sherpa-onnx-streaming-zipformer-de-kroko-2025-08-06",
|
1609 |
]:
|
1610 |
encoder_model = _get_nn_model_filename(
|
1611 |
repo_id=repo_id,
|
|
|
2095 |
"csukuangfj/wenet-english-model": _get_wenet_model,
|
2096 |
}
|
2097 |
|
2098 |
+
twenty_five_languages_models = {
|
2099 |
+
"csukuangfj/sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8": _get_sherpa_onnx_nemo_transducer_models_int8,
|
2100 |
+
}
|
2101 |
+
|
2102 |
multi_lingual_models = {
|
2103 |
"csukuangfj/sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02": _get_dolphin_ctc_models,
|
2104 |
"csukuangfj/sherpa-onnx-dolphin-small-ctc-multi-lang-int8-2025-04-02": _get_dolphin_ctc_models,
|
|
|
2125 |
}
|
2126 |
|
2127 |
german_models = {
|
2128 |
+
"csukuangfj/sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8": _get_sherpa_onnx_nemo_transducer_models_int8,
|
2129 |
+
"csukuangfj/sherpa-onnx-streaming-zipformer-de-kroko-2025-08-06": _get_streaming_zipformer_pre_trained_model,
|
2130 |
"csukuangfj/sherpa-onnx-nemo-transducer-stt_de_fastconformer_hybrid_large_pc": _get_sherpa_onnx_nemo_transducer_models,
|
2131 |
"csukuangfj/sherpa-onnx-nemo-transducer-stt_de_fastconformer_hybrid_large_pc-int8": _get_sherpa_onnx_nemo_transducer_models_int8,
|
2132 |
"csukuangfj/sherpa-onnx-nemo-stt_de_fastconformer_hybrid_large_pc": _get_sherpa_onnx_nemo_ctc_models,
|
|
|
2135 |
}
|
2136 |
|
2137 |
french_models = {
|
2138 |
+
"csukuangfj/sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8": _get_sherpa_onnx_nemo_transducer_models_int8,
|
2139 |
"shaojieli/sherpa-onnx-streaming-zipformer-fr-2023-04-14": _get_french_pre_trained_model,
|
2140 |
"csukuangfj/sherpa-onnx-streaming-zipformer-fr-kroko-2025-08-06": _get_streaming_zipformer_pre_trained_model,
|
2141 |
}
|
2142 |
|
2143 |
spanish_models = {
|
2144 |
+
"csukuangfj/sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8": _get_sherpa_onnx_nemo_transducer_models_int8,
|
2145 |
"csukuangfj/sherpa-onnx-streaming-zipformer-es-kroko-2025-08-06": _get_streaming_zipformer_pre_trained_model,
|
2146 |
}
|
2147 |
|
|
|
2152 |
}
|
2153 |
|
2154 |
russian_models = {
|
2155 |
+
"csukuangfj/sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8": _get_sherpa_onnx_nemo_transducer_models_int8,
|
2156 |
"csukuangfj/sherpa-onnx-nemo-transducer-giga-am-v2-russian-2025-04-19": _get_russian_pre_trained_model,
|
2157 |
"csukuangfj/sherpa-onnx-nemo-ctc-giga-am-v2-russian-2025-04-19": _get_russian_pre_trained_model_ctc,
|
2158 |
"csukuangfj/sherpa-onnx-nemo-transducer-giga-am-russian-2024-10-24": _get_russian_pre_trained_model,
|
|
|
2189 |
}
|
2190 |
|
2191 |
portuguese_brazlian_models = {
|
2192 |
+
"csukuangfj/sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8": _get_sherpa_onnx_nemo_transducer_models_int8,
|
2193 |
"csukuangfj/sherpa-onnx-nemo-stt_pt_fastconformer_hybrid_large_pc": _get_sherpa_onnx_nemo_ctc_models,
|
2194 |
"csukuangfj/sherpa-onnx-nemo-stt_pt_fastconformer_hybrid_large_pc-int8": _get_sherpa_onnx_nemo_ctc_models,
|
2195 |
"csukuangfj/sherpa-onnx-nemo-transducer-stt_pt_fastconformer_hybrid_large_pc": _get_sherpa_onnx_nemo_transducer_models,
|
|
|
2198 |
|
2199 |
|
2200 |
all_models = {
|
2201 |
+
**twenty_five_languages_models,
|
2202 |
**multi_lingual_models,
|
2203 |
**chinese_models,
|
2204 |
**english_models,
|
|
|
2220 |
}
|
2221 |
|
2222 |
language_to_models = {
|
2223 |
+
"25 European languages": list(twenty_five_languages_models.keys()),
|
2224 |
"Multi-lingual (east aisa)": list(multi_lingual_models.keys()),
|
2225 |
"超多种中文方言": list(chinese_dialect_models.keys()),
|
2226 |
"Chinese": list(chinese_models.keys()),
|