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

Add C++ runtime and Python APIs for Moonshine models (#1473)

Browse Source
This commit is contained in:
Fangjun Kuang
2024-10-26 14:34:07 +08:00
committed by GitHub
parent 0f2732e4e8
commit 669f5ef441
33 changed files with 1572 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

@@ -11,6 +11,7 @@ set(srcs
offline-ctc-fst-decoder-config.cc
offline-lm-config.cc
offline-model-config.cc
offline-moonshine-model-config.cc
offline-nemo-enc-dec-ctc-model-config.cc
offline-paraformer-model-config.cc
offline-punctuation.cc

7
sherpa-onnx/python/csrc/offline-model-config.cc
View File

@@ -8,6 +8,7 @@
#include <vector>
#include "sherpa-onnx/csrc/offline-model-config.h"
#include "sherpa-onnx/python/csrc/offline-moonshine-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-sense-voice-model-config.h"
@@ -28,6 +29,7 @@ void PybindOfflineModelConfig(py::module *m) {
PybindOfflineZipformerCtcModelConfig(m);
PybindOfflineWenetCtcModelConfig(m);
PybindOfflineSenseVoiceModelConfig(m);
PybindOfflineMoonshineModelConfig(m);
using PyClass = OfflineModelConfig;
py::class_<PyClass>(*m, "OfflineModelConfig")
@@ -39,7 +41,8 @@ void PybindOfflineModelConfig(py::module *m) {
const OfflineWhisperModelConfig &, const OfflineTdnnModelConfig &,
const OfflineZipformerCtcModelConfig &,
const OfflineWenetCtcModelConfig &,
const OfflineSenseVoiceModelConfig &, const std::string &,
const OfflineSenseVoiceModelConfig &,
const OfflineMoonshineModelConfig &, const std::string &,
const std::string &, int32_t, bool, const std::string &,
const std::string &, const std::string &, const std::string &>(),
py::arg("transducer") = OfflineTransducerModelConfig(),
@@ -50,6 +53,7 @@ void PybindOfflineModelConfig(py::module *m) {
py::arg("zipformer_ctc") = OfflineZipformerCtcModelConfig(),
py::arg("wenet_ctc") = OfflineWenetCtcModelConfig(),
py::arg("sense_voice") = OfflineSenseVoiceModelConfig(),
py::arg("moonshine") = OfflineMoonshineModelConfig(),
py::arg("telespeech_ctc") = "", py::arg("tokens"),
py::arg("num_threads"), py::arg("debug") = false,
py::arg("provider") = "cpu", py::arg("model_type") = "",
@@ -62,6 +66,7 @@ void PybindOfflineModelConfig(py::module *m) {
.def_readwrite("zipformer_ctc", &PyClass::zipformer_ctc)
.def_readwrite("wenet_ctc", &PyClass::wenet_ctc)
.def_readwrite("sense_voice", &PyClass::sense_voice)
.def_readwrite("moonshine", &PyClass::moonshine)
.def_readwrite("telespeech_ctc", &PyClass::telespeech_ctc)
.def_readwrite("tokens", &PyClass::tokens)
.def_readwrite("num_threads", &PyClass::num_threads)

28
sherpa-onnx/python/csrc/offline-moonshine-model-config.cc Normal file
View File

@@ -0,0 +1,28 @@
// sherpa-onnx/python/csrc/offline-moonshine-model-config.cc
//
// Copyright (c) 2024 Xiaomi Corporation
#include "sherpa-onnx/csrc/offline-moonshine-model-config.h"
#include <string>
#include <vector>
#include "sherpa-onnx/python/csrc/offline-moonshine-model-config.h"
namespace sherpa_onnx {
void PybindOfflineMoonshineModelConfig(py::module *m) {
using PyClass = OfflineMoonshineModelConfig;
py::class_<PyClass>(*m, "OfflineMoonshineModelConfig")
.def(py::init<const std::string &, const std::string &,
const std::string &, const std::string &>(),
py::arg("preprocessor"), py::arg("encoder"),
py::arg("uncached_decoder"), py::arg("cached_decoder"))
.def_readwrite("preprocessor", &PyClass::preprocessor)
.def_readwrite("encoder", &PyClass::encoder)
.def_readwrite("uncached_decoder", &PyClass::uncached_decoder)
.def_readwrite("cached_decoder", &PyClass::cached_decoder)
.def("__str__", &PyClass::ToString);
}
} // namespace sherpa_onnx

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

@@ -0,0 +1,16 @@
// sherpa-onnx/python/csrc/offline-moonshine-model-config.h
//
// Copyright (c) 2024 Xiaomi Corporation
#ifndef SHERPA_ONNX_PYTHON_CSRC_OFFLINE_MOONSHINE_MODEL_CONFIG_H_
#define SHERPA_ONNX_PYTHON_CSRC_OFFLINE_MOONSHINE_MODEL_CONFIG_H_
#include "sherpa-onnx/python/csrc/sherpa-onnx.h"
namespace sherpa_onnx {
void PybindOfflineMoonshineModelConfig(py::module *m);
}
#endif // SHERPA_ONNX_PYTHON_CSRC_OFFLINE_MOONSHINE_MODEL_CONFIG_H_

92
sherpa-onnx/python/sherpa_onnx/offline_recognizer.py
View File

@@ -8,13 +8,14 @@ from _sherpa_onnx import (
OfflineCtcFstDecoderConfig,
OfflineLMConfig,
OfflineModelConfig,
OfflineMoonshineModelConfig,
OfflineNemoEncDecCtcModelConfig,
OfflineParaformerModelConfig,
OfflineSenseVoiceModelConfig,
)
from _sherpa_onnx import OfflineRecognizer as _Recognizer
from _sherpa_onnx import (
OfflineRecognizerConfig,
OfflineSenseVoiceModelConfig,
OfflineStream,
OfflineTdnnModelConfig,
OfflineTransducerModelConfig,
@@ -503,12 +504,12 @@ class OfflineRecognizer(object):
e.g., tiny, tiny.en, base, base.en, etc.
Args:
encoder_model:
Path to the encoder model, e.g., tiny-encoder.onnx,
tiny-encoder.int8.onnx, tiny-encoder.ort, etc.
decoder_model:
encoder:
Path to the encoder model, e.g., tiny-encoder.onnx,
tiny-encoder.int8.onnx, tiny-encoder.ort, etc.
decoder:
Path to the decoder model, e.g., tiny-decoder.onnx,
tiny-decoder.int8.onnx, tiny-decoder.ort, etc.
tokens:
Path to ``tokens.txt``. Each line in ``tokens.txt`` contains two
columns::
@@ -570,6 +571,87 @@ class OfflineRecognizer(object):
self.config = recognizer_config
return self
@classmethod
def from_moonshine(
cls,
preprocessor: str,
encoder: str,
uncached_decoder: str,
cached_decoder: str,
tokens: str,
num_threads: int = 1,
decoding_method: str = "greedy_search",
debug: bool = False,
provider: str = "cpu",
rule_fsts: str = "",
rule_fars: str = "",
):
"""
Please refer to
`<https://k2-fsa.github.io/sherpa/onnx/moonshine/index.html>`_
to download pre-trained models for different kinds of moonshine models,
e.g., tiny, base, etc.
Args:
preprocessor:
Path to the preprocessor model, e.g., preprocess.onnx
encoder:
Path to the encoder model, e.g., encode.int8.onnx
uncached_decoder:
Path to the uncached decoder model, e.g., uncached_decode.int8.onnx,
cached_decoder:
Path to the cached decoder model, e.g., cached_decode.int8.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.
decoding_method:
Valid values: greedy_search.
debug:
True to show debug messages.
provider:
onnxruntime execution providers. Valid values are: cpu, cuda, coreml.
rule_fsts:
If not empty, it specifies fsts for inverse text normalization.
If there are multiple fsts, they are separated by a comma.
rule_fars:
If not empty, it specifies fst archives for inverse text normalization.
If there are multiple archives, they are separated by a comma.
"""
self = cls.__new__(cls)
model_config = OfflineModelConfig(
moonshine=OfflineMoonshineModelConfig(
preprocessor=preprocessor,
encoder=encoder,
uncached_decoder=uncached_decoder,
cached_decoder=cached_decoder,
),
tokens=tokens,
num_threads=num_threads,
debug=debug,
provider=provider,
)
unused_feat_config = FeatureExtractorConfig(
sampling_rate=16000,
feature_dim=80,
)
recognizer_config = OfflineRecognizerConfig(
model_config=model_config,
feat_config=unused_feat_config,
decoding_method=decoding_method,
rule_fsts=rule_fsts,
rule_fars=rule_fars,
)
self.recognizer = _Recognizer(recognizer_config)
self.config = recognizer_config
return self
@classmethod
def from_tdnn_ctc(
cls,