Skip to content

Commit 6f513ca

Browse files
puneithkpuneithk
committed
added async
Author: puneith <puneith@google.com>
1 parent 20e6b88 commit 6f513ca

File tree

1 file changed

+219
-0
lines changed

1 file changed

+219
-0
lines changed

‎speech/grpc/src/main/java/com/google/cloud/speech/grpc/demos/AsyncRecognizeClient.java

Lines changed: 219 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,219 @@
1+
/*
2+
* Copyright 2016 Google Inc. All Rights Reserved.
3+
*
4+
* Licensed under the Apache License, Version 2.0 (the "License");
5+
* you may not use this file except in compliance with the License.
6+
* You may obtain a copy of the License at
7+
*
8+
* http://www.apache.org/licenses/LICENSE-2.0
9+
*
10+
* Unless required by applicable law or agreed to in writing, software
11+
* distributed under the License is distributed on an "AS IS" BASIS,
12+
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13+
* See the License for the specific language governing permissions and
14+
* limitations under the License.
15+
*/
16+
17+
// Client that sends audio to Speech.AsyncRecognize via gRPC and returns transcription.
18+
//
19+
// Uses a service account for OAuth2 authentication, which you may obtain at
20+
// https://console.developers.google.com
21+
// API Manager > Google Cloud Speech API > Enable
22+
// API Manager > Credentials > Create credentials > Service account key > New service account.
23+
//
24+
// Then set environment variable GOOGLE_APPLICATION_CREDENTIALS to the full path of that file.
25+
26+
package com.google.cloud.speech.grpc.demos;
27+
28+
import com.google.auth.oauth2.GoogleCredentials;
29+
import com.google.cloud.speech.v1beta1.RecognitionAudio;
30+
import com.google.cloud.speech.v1beta1.RecognitionConfig;
31+
import com.google.cloud.speech.v1beta1.RecognitionConfig.AudioEncoding;
32+
import com.google.cloud.speech.v1beta1.SpeechGrpc;
33+
import com.google.cloud.speech.v1beta1.AsyncRecognizeRequest;
34+
import com.google.cloud.speech.v1beta1.AsyncRecognizeResponse;
35+
import com.google.protobuf.TextFormat;
36+
37+
import google.longrunning.Operation;
38+
39+
import io.grpc.ManagedChannel;
40+
import io.grpc.StatusRuntimeException;
41+
import io.grpc.auth.ClientAuthInterceptor;
42+
import io.grpc.netty.NegotiationType;
43+
import io.grpc.netty.NettyChannelBuilder;
44+
45+
import org.apache.commons.cli.CommandLine;
46+
import org.apache.commons.cli.CommandLineParser;
47+
import org.apache.commons.cli.DefaultParser;
48+
import org.apache.commons.cli.OptionBuilder;
49+
import org.apache.commons.cli.Options;
50+
import org.apache.commons.cli.ParseException;
51+
52+
import java.io.IOException;
53+
import java.net.URI;
54+
import java.util.Arrays;
55+
import java.util.List;
56+
import java.util.concurrent.Executors;
57+
import java.util.concurrent.TimeUnit;
58+
import java.util.logging.Level;
59+
import java.util.logging.Logger;
60+
61+
/**
62+
* Client that sends audio to Speech.AsyncRecognize and returns transcript.
63+
*/
64+
public class AsyncRecognizeClient {
65+
66+
private static final Logger logger =
67+
Logger.getLogger(AsyncRecognizeClient.class.getName());
68+
69+
private static final List<String> OAUTH2_SCOPES =
70+
Arrays.asList("https://www.googleapis.com/auth/cloud-platform");
71+
72+
private final String host;
73+
private final int port;
74+
private final URI input;
75+
private final int samplingRate;
76+
77+
private final ManagedChannel channel;
78+
private final SpeechGrpc.SpeechBlockingStub stub;
79+
80+
/**
81+
* Construct client connecting to Cloud Speech server at {@code host:port}.
82+
*/
83+
public AsyncRecognizeClient(String host, int port, URI input, int samplingRate)
84+
throws IOException {
85+
this.host = host;
86+
this.port = port;
87+
this.input = input;
88+
this.samplingRate = samplingRate;
89+
90+
GoogleCredentials creds = GoogleCredentials.getApplicationDefault();
91+
creds = creds.createScoped(OAUTH2_SCOPES);
92+
channel = NettyChannelBuilder.forAddress(host, port)
93+
.negotiationType(NegotiationType.TLS)
94+
.intercept(new ClientAuthInterceptor(creds, Executors.newSingleThreadExecutor()))
95+
.build();
96+
stub = SpeechGrpc.newBlockingStub(channel);
97+
logger.info("Created stub for " + host + ":" + port);
98+
}
99+
100+
private RecognitionAudio createRecognitionAudio() throws IOException {
101+
return RecognitionAudioFactory.createRecognitionAudio(this.input);
102+
}
103+
104+
public void shutdown() throws InterruptedException {
105+
channel.shutdown().awaitTermination(5, TimeUnit.SECONDS);
106+
}
107+
108+
/** Send an async-recognize request to server. */
109+
public void recognize() {
110+
RecognitionAudio audio;
111+
try {
112+
audio = createRecognitionAudio();
113+
} catch (IOException e) {
114+
logger.log(Level.WARNING, "Failed to read audio uri input: " + input);
115+
return;
116+
}
117+
logger.info("Sending " + audio.getContent().size() + " bytes from audio uri input: " + input);
118+
RecognitionConfig config = RecognitionConfig.newBuilder()
119+
.setEncoding(AudioEncoding.LINEAR16)
120+
.setSampleRate(samplingRate)
121+
.build();
122+
AsyncRecognizeRequest request = AsyncRecognizeRequest.newBuilder()
123+
.setConfig(config)
124+
.setAudio(audio)
125+
.build();
126+
127+
Operation operation;
128+
try {
129+
operation = stub.asyncRecognize(request);
130+
131+
//Print the long running operation handle
132+
System.out.println(operation);
133+
} catch (StatusRuntimeException e) {
134+
logger.log(Level.WARNING, "RPC failed: {0}", e.getStatus());
135+
return;
136+
}
137+
138+
while(operation.hasDone()) {
139+
System.out.println(operation);
140+
//Sleep for 1 s
141+
Thread.sleep(1000);
142+
}
143+
//logger.info("Received response: " + TextFormat.printToString(response));
144+
}
145+
146+
public static void main(String[] args) throws Exception {
147+
148+
String audioFile = "";
149+
String host = "speech.googleapis.com";
150+
Integer port = 443;
151+
Integer sampling = 16000;
152+
153+
CommandLineParser parser = new DefaultParser();
154+
155+
Options options = new Options();
156+
options.addOption(OptionBuilder.withLongOpt("uri")
157+
.withDescription("path to audio uri")
158+
.hasArg()
159+
.withArgName("FILE_PATH")
160+
.create());
161+
options.addOption(OptionBuilder.withLongOpt("host")
162+
.withDescription("endpoint for api, e.g. speech.googleapis.com")
163+
.hasArg()
164+
.withArgName("ENDPOINT")
165+
.create());
166+
options.addOption(OptionBuilder.withLongOpt("port")
167+
.withDescription("SSL port, usually 443")
168+
.hasArg()
169+
.withArgName("PORT")
170+
.create());
171+
options.addOption(OptionBuilder.withLongOpt("sampling")
172+
.withDescription("Sampling Rate, i.e. 16000")
173+
.hasArg()
174+
.withArgName("RATE")
175+
.create());
176+
177+
try {
178+
CommandLine line = parser.parse(options, args);
179+
if (line.hasOption("uri")) {
180+
audioFile = line.getOptionValue("uri");
181+
} else {
182+
System.err.println("An Audio uri must be specified (e.g. file:///foo/baz.raw).");
183+
System.exit(1);
184+
}
185+
186+
if (line.hasOption("host")) {
187+
host = line.getOptionValue("host");
188+
} else {
189+
System.err.println("An API enpoint must be specified (typically speech.googleapis.com).");
190+
System.exit(1);
191+
}
192+
193+
if (line.hasOption("port")) {
194+
port = Integer.parseInt(line.getOptionValue("port"));
195+
} else {
196+
System.err.println("An SSL port must be specified (typically 443).");
197+
System.exit(1);
198+
}
199+
200+
if (line.hasOption("sampling")) {
201+
sampling = Integer.parseInt(line.getOptionValue("sampling"));
202+
} else {
203+
System.err.println("An Audio sampling rate must be specified.");
204+
System.exit(1);
205+
}
206+
} catch (ParseException exp) {
207+
System.err.println("Unexpected exception:" + exp.getMessage());
208+
System.exit(1);
209+
}
210+
211+
AsyncRecognizeClient client =
212+
new AsyncRecognizeClient(host, port, URI.create(audioFile), sampling);
213+
try {
214+
client.recognize();
215+
} finally {
216+
client.shutdown();
217+
}
218+
}
219+
}

0 commit comments

Comments
 (0)