online-websocket-server.cc
2.8 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
// sherpa-onnx/csrc/online-websocket-server.cc
//
// Copyright (c) 2022-2023 Xiaomi Corporation
#include "asio.hpp"
#include "sherpa-onnx/csrc/macros.h"
#include "sherpa-onnx/csrc/online-websocket-server-impl.h"
#include "sherpa-onnx/csrc/parse-options.h"
static constexpr const char *kUsageMessage = R"(
Automatic speech recognition with sherpa-onnx using websocket.
Usage:
./bin/sherpa-onnx-online-websocket-server --help
./bin/sherpa-onnx-online-websocket-server \
--port=6006 \
--num-work-threads=5 \
--tokens=/path/to/tokens.txt \
--encoder=/path/to/encoder.onnx \
--decoder=/path/to/decoder.onnx \
--joiner=/path/to/joiner.onnx \
--log-file=./log.txt \
--max-batch-size=5 \
--loop-interval-ms=10
Please refer to
https://k2-fsa.github.io/sherpa/onnx/pretrained_models/index.html
for a list of pre-trained models to download.
)";
int32_t main(int32_t argc, char *argv[]) {
sherpa_onnx::ParseOptions po(kUsageMessage);
sherpa_onnx::OnlineWebsocketServerConfig config;
// the server will listen on this port
int32_t port = 6006;
// size of the thread pool for handling network connections
int32_t num_io_threads = 1;
// size of the thread pool for neural network computation and decoding
int32_t num_work_threads = 3;
po.Register("num-io-threads", &num_io_threads,
"Thread pool size for network connections.");
po.Register("num-work-threads", &num_work_threads,
"Thread pool size for for neural network "
"computation and decoding.");
po.Register("port", &port, "The port on which the server will listen.");
config.Register(&po);
if (argc == 1) {
po.PrintUsage();
exit(EXIT_FAILURE);
}
po.Read(argc, argv);
if (po.NumArgs() != 0) {
SHERPA_ONNX_LOGE("Unrecognized positional arguments!");
po.PrintUsage();
exit(EXIT_FAILURE);
}
config.Validate();
asio::io_context io_conn; // for network connections
asio::io_context io_work; // for neural network and decoding
sherpa_onnx::OnlineWebsocketServer server(io_conn, io_work, config);
server.Run(port);
SHERPA_ONNX_LOGE("Listening on: %d", port);
SHERPA_ONNX_LOGE("Number of work threads: %d", num_work_threads);
// give some work to do for the io_work pool
auto work_guard = asio::make_work_guard(io_work);
std::vector<std::thread> io_threads;
// decrement since the main thread is also used for network communications
for (int32_t i = 0; i < num_io_threads - 1; ++i) {
io_threads.emplace_back([&io_conn]() { io_conn.run(); });
}
std::vector<std::thread> work_threads;
for (int32_t i = 0; i < num_work_threads; ++i) {
work_threads.emplace_back([&io_work]() { io_work.run(); });
}
io_conn.run();
for (auto &t : io_threads) {
t.join();
}
for (auto &t : work_threads) {
t.join();
}
return 0;
}