EngineX-Iluvatar/enginex_bi_series-sherpa-onnx
Archived
16
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

Support TDNN models from the yesno recipe from icefall (#262)

Browse Source
This commit is contained in:
Fangjun Kuang
2023-08-12 19:50:22 +08:00
committed by GitHub
parent b094868fb8
commit a4bff28e21
23 changed files with 612 additions and 36 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
sherpa-onnx/python/csrc/CMakeLists.txt
View File

@@ -10,6 +10,7 @@ pybind11_add_module(_sherpa_onnx
offline-paraformer-model-config.cc
offline-recognizer.cc
offline-stream.cc
offline-tdnn-model-config.cc
offline-transducer-model-config.cc
offline-whisper-model-config.cc
online-lm-config.cc

9
sherpa-onnx/python/csrc/offline-model-config.cc
View File

@@ -10,6 +10,7 @@
#include "sherpa-onnx/csrc/offline-model-config.h"
#include "sherpa-onnx/python/csrc/offline-nemo-enc-dec-ctc-model-config.h"
#include "sherpa-onnx/python/csrc/offline-paraformer-model-config.h"
#include "sherpa-onnx/python/csrc/offline-tdnn-model-config.h"
#include "sherpa-onnx/python/csrc/offline-transducer-model-config.h"
#include "sherpa-onnx/python/csrc/offline-whisper-model-config.h"
@@ -20,24 +21,28 @@ void PybindOfflineModelConfig(py::module *m) {
PybindOfflineParaformerModelConfig(m);
PybindOfflineNemoEncDecCtcModelConfig(m);
PybindOfflineWhisperModelConfig(m);
PybindOfflineTdnnModelConfig(m);
using PyClass = OfflineModelConfig;
py::class_<PyClass>(*m, "OfflineModelConfig")
.def(py::init<const OfflineTransducerModelConfig &,
const OfflineParaformerModelConfig &,
const OfflineNemoEncDecCtcModelConfig &,
const OfflineWhisperModelConfig &, const std::string &,
const OfflineWhisperModelConfig &,
const OfflineTdnnModelConfig &, const std::string &,
int32_t, bool, const std::string &, const std::string &>(),
py::arg("transducer") = OfflineTransducerModelConfig(),
py::arg("paraformer") = OfflineParaformerModelConfig(),
py::arg("nemo_ctc") = OfflineNemoEncDecCtcModelConfig(),
py::arg("whisper") = OfflineWhisperModelConfig(), py::arg("tokens"),
py::arg("whisper") = OfflineWhisperModelConfig(),
py::arg("tdnn") = OfflineTdnnModelConfig(), py::arg("tokens"),
py::arg("num_threads"), py::arg("debug") = false,
py::arg("provider") = "cpu", py::arg("model_type") = "")
.def_readwrite("transducer", &PyClass::transducer)
.def_readwrite("paraformer", &PyClass::paraformer)
.def_readwrite("nemo_ctc", &PyClass::nemo_ctc)
.def_readwrite("whisper", &PyClass::whisper)
.def_readwrite("tdnn", &PyClass::tdnn)
.def_readwrite("tokens", &PyClass::tokens)
.def_readwrite("num_threads", &PyClass::num_threads)
.def_readwrite("debug", &PyClass::debug)

22
sherpa-onnx/python/csrc/offline-tdnn-model-config.cc Normal file
View File

@@ -0,0 +1,22 @@
// sherpa-onnx/python/csrc/offline-tdnn-model-config.cc
//
// Copyright (c) 2023 Xiaomi Corporation
#include "sherpa-onnx/csrc/offline-tdnn-model-config.h"
#include <string>
#include <vector>
#include "sherpa-onnx/python/csrc/offline-tdnn-model-config.h"
namespace sherpa_onnx {
void PybindOfflineTdnnModelConfig(py::module *m) {
using PyClass = OfflineTdnnModelConfig;
py::class_<PyClass>(*m, "OfflineTdnnModelConfig")
.def(py::init<const std::string &>(), py::arg("model"))
.def_readwrite("model", &PyClass::model)
.def("__str__", &PyClass::ToString);
}
} // namespace sherpa_onnx

16
sherpa-onnx/python/csrc/offline-tdnn-model-config.h Normal file
View File

@@ -0,0 +1,16 @@
// sherpa-onnx/python/csrc/offline-tdnn-model-config.h
//
// Copyright (c) 2023 Xiaomi Corporation
#ifndef SHERPA_ONNX_PYTHON_CSRC_OFFLINE_TDNN_MODEL_CONFIG_H_
#define SHERPA_ONNX_PYTHON_CSRC_OFFLINE_TDNN_MODEL_CONFIG_H_
#include "sherpa-onnx/python/csrc/sherpa-onnx.h"
namespace sherpa_onnx {
void PybindOfflineTdnnModelConfig(py::module *m);
}
#endif // SHERPA_ONNX_PYTHON_CSRC_OFFLINE_TDNN_MODEL_CONFIG_H_

83
sherpa-onnx/python/sherpa_onnx/offline_recognizer.py
View File

@@ -8,6 +8,7 @@ from _sherpa_onnx import (
OfflineModelConfig,
OfflineNemoEncDecCtcModelConfig,
OfflineParaformerModelConfig,
OfflineTdnnModelConfig,
OfflineWhisperModelConfig,
)
from _sherpa_onnx import OfflineRecognizer as _Recognizer
@@ -37,7 +38,7 @@ class OfflineRecognizer(object):
decoder: str,
joiner: str,
tokens: str,
num_threads: int,
num_threads: int = 1,
sample_rate: int = 16000,
feature_dim: int = 80,
decoding_method: str = "greedy_search",
@@ -48,7 +49,7 @@ class OfflineRecognizer(object):
):
"""
Please refer to
`<https://k2-fsa.github.io/sherpa/onnx/pretrained_models/index.html>`_
`<https://k2-fsa.github.io/sherpa/onnx/pretrained_models/offline-transducer/index.html>`_
to download pre-trained models for different languages, e.g., Chinese,
English, etc.
@@ -115,7 +116,7 @@ class OfflineRecognizer(object):
cls,
paraformer: str,
tokens: str,
num_threads: int,
num_threads: int = 1,
sample_rate: int = 16000,
feature_dim: int = 80,
decoding_method: str = "greedy_search",
@@ -124,9 +125,8 @@ class OfflineRecognizer(object):
):
"""
Please refer to
`<https://k2-fsa.github.io/sherpa/onnx/pretrained_models/index.html>`_
to download pre-trained models for different languages, e.g., Chinese,
English, etc.
`<https://k2-fsa.github.io/sherpa/onnx/pretrained_models/offline-paraformer/index.html>`_
to download pre-trained models.
Args:
tokens:
@@ -179,7 +179,7 @@ class OfflineRecognizer(object):
cls,
model: str,
tokens: str,
num_threads: int,
num_threads: int = 1,
sample_rate: int = 16000,
feature_dim: int = 80,
decoding_method: str = "greedy_search",
@@ -188,7 +188,7 @@ class OfflineRecognizer(object):
):
"""
Please refer to
`<https://k2-fsa.github.io/sherpa/onnx/pretrained_models/index.html>`_
`<https://k2-fsa.github.io/sherpa/onnx/pretrained_models/offline-ctc/nemo/index.html>`_
to download pre-trained models for different languages, e.g., Chinese,
English, etc.
@@ -244,14 +244,14 @@ class OfflineRecognizer(object):
encoder: str,
decoder: str,
tokens: str,
num_threads: int,
num_threads: int = 1,
decoding_method: str = "greedy_search",
debug: bool = False,
provider: str = "cpu",
):
"""
Please refer to
`<https://k2-fsa.github.io/sherpa/onnx/pretrained_models/index.html>`_
`<https://k2-fsa.github.io/sherpa/onnx/pretrained_models/whisper/index.html>`_
to download pre-trained models for different kinds of whisper models,
e.g., tiny, tiny.en, base, base.en, etc.
@@ -301,6 +301,69 @@ class OfflineRecognizer(object):
self.config = recognizer_config
return self
@classmethod
def from_tdnn_ctc(
cls,
model: str,
tokens: str,
num_threads: int = 1,
sample_rate: int = 8000,
feature_dim: int = 23,
decoding_method: str = "greedy_search",
debug: bool = False,
provider: str = "cpu",
):
"""
Please refer to
`<https://k2-fsa.github.io/sherpa/onnx/pretrained_models/offline-ctc/yesno/index.html>`_
to download pre-trained models.
Args:
model:
Path to ``model.onnx``.
tokens:
Path to ``tokens.txt``. Each line in ``tokens.txt`` contains two
columns::
symbol integer_id
num_threads:
Number of threads for neural network computation.
sample_rate:
Sample rate of the training data used to train the model.
feature_dim:
Dimension of the feature used to train the model.
decoding_method:
Valid values are greedy_search.
debug:
True to show debug messages.
provider:
onnxruntime execution providers. Valid values are: cpu, cuda, coreml.
"""
self = cls.__new__(cls)
model_config = OfflineModelConfig(
tdnn=OfflineTdnnModelConfig(model=model),
tokens=tokens,
num_threads=num_threads,
debug=debug,
provider=provider,
model_type="tdnn",
)
feat_config = OfflineFeatureExtractorConfig(
sampling_rate=sample_rate,
feature_dim=feature_dim,
)
recognizer_config = OfflineRecognizerConfig(
feat_config=feat_config,
model_config=model_config,
decoding_method=decoding_method,
)
self.recognizer = _Recognizer(recognizer_config)
self.config = recognizer_config
return self
def create_stream(self, contexts_list: Optional[List[List[int]]] = None):
if contexts_list is None:
return self.recognizer.create_stream()