All Downloads are FREE. Search and download functionalities are using the official Maven repository.

com.alibaba.nls.client.protocol.tts.SpeechSynthesizer Maven / Gradle / Ivy

There is a newer version: 2.2.16
Show newest version
/*
 * Copyright 2015 Alibaba Group Holding Limited
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

package com.alibaba.nls.client.protocol.tts;

import java.util.HashMap;
import java.util.concurrent.CountDownLatch;
import java.util.concurrent.TimeUnit;

import com.alibaba.nls.client.protocol.Constant;
import com.alibaba.nls.client.protocol.NlsClient;
import com.alibaba.nls.client.protocol.OutputFormatEnum;
import com.alibaba.nls.client.protocol.SampleRateEnum;
import com.alibaba.nls.client.protocol.SpeechReqProtocol;

import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

import static com.alibaba.nls.client.protocol.SpeechReqProtocol.State.STATE_CLOSED;
import static com.alibaba.nls.client.protocol.SpeechReqProtocol.State.STATE_REQUEST_CONFIRMED;

/**
 * @author zhishen.ml
 * @date 2017/11/28
 *
 * 语音合成
 */
public class SpeechSynthesizer extends SpeechReqProtocol {
    static Logger logger = LoggerFactory.getLogger(SpeechSynthesizer.class);
    private SpeechSynthesizerListener listener;
    private CountDownLatch completeLatch;
    private boolean isLongText;

    /**
     * 如果没有设置format,默认为pcm
     */
    private static final String DEFAULT_FORMAT = "pcm";
    /**
     * 如果没有设置sampleRate,默认为16000
     */
    private static final Integer DEFAULT_SAMPLE_RATE = 16000;

    private static final Integer DEFAULT_VOICE_VOLUME = 50;

    public SpeechSynthesizer(NlsClient client, SpeechSynthesizerListener listener) throws Exception{
        conn = client.connect(listener);
        this.listener = listener;
        afterConnection(listener);
    }

    public SpeechSynthesizer(NlsClient client,String token, SpeechSynthesizerListener listener) throws Exception{
        this.conn = client.connect(token,listener);
        this.listener = listener;
        afterConnection(listener);
    }

    protected void afterConnection(SpeechSynthesizerListener listener){
        payload = new HashMap();
        header.put(Constant.PROP_NAMESPACE, TTSConstant.VALUE_NAMESPACE_TTS);
        header.put(Constant.PROP_NAME, TTSConstant.VALUE_NAME_TTS_START);
        payload.put(TTSConstant.PROP_TTS_FORMAT, DEFAULT_FORMAT);
        payload.put(TTSConstant.PROP_TTS_SAMPLE_RATE, DEFAULT_SAMPLE_RATE);
        payload.put(TTSConstant.PROP_TTS_VOLUME, DEFAULT_VOICE_VOLUME);
        listener.setSpeechSynthesizer(this);
        state = State.STATE_CONNECTED;
    }

    /**
     * 发音人
     *
     * @param voice
     */
    public void setVoice(String voice) {
        payload.put(TTSConstant.PROP_TTS_VOICE, voice);
    }

    /**
     * 待合成文本
     *
     * @param text
     */
    public void setText(String text) {
        isLongText = false;
        header.put(Constant.PROP_NAMESPACE, TTSConstant.VALUE_NAMESPACE_TTS);
        payload.put(TTSConstant.PROP_TTS_TEXT, text);
    }

    public void setLongText(String text) {
        isLongText = true;
        header.put(Constant.PROP_NAMESPACE, TTSConstant.VALUE_NAMESPACE_LONG_TTS);
        payload.put(TTSConstant.PROP_TTS_TEXT, text);
    }

    /**
     * 合成语音的编码格式,支持的格式:pcm,wav,mp3 默认是pcm
     *
     * @param format
     */
    public void setFormat(OutputFormatEnum format) {
        payload.put(Constant.PROP_ASR_FORMAT, format.getName());
    }

    /**
     * 合成语音的采样率,可选默认16000
     *
     * @param sampleRate
     */
    public void setSampleRate(SampleRateEnum sampleRate) {
        payload.put(TTSConstant.PROP_TTS_SAMPLE_RATE, sampleRate.value);
    }

    /**
     * 个别场景需要设置除'SampleRateEnum'之外的采样率,如32000,采用此方法进行设置
     * @param sampleRate
     */
    public void setSampleRate(int sampleRate) {
        payload.put(TTSConstant.PROP_TTS_SAMPLE_RATE, sampleRate);
    }

    /**
     * 音量,范围是0~100,可选,默认50
     *
     * @param volume
     */
    public void setVolume(int volume) {
        payload.put(TTSConstant.PROP_TTS_VOLUME, volume);
    }

    /**
     * 语速,范围是-500~500,可选,默认是0
     *
     * @param speechRate
     */
    public void setSpeechRate(int speechRate) {
        payload.put(TTSConstant.PROP_TTS_SPEECH_RATE, speechRate);
    }

    /**
     * 语调,范围是-500~500,可选,默认是0
     *
     * @param pitchRate
     */
    public void setPitchRate(int pitchRate) {
        payload.put(TTSConstant.PROP_TTS_PITCH_RATE, pitchRate);
    }

    /**
     * 语音合成方法 默认为0
     *
     * 0 统计参数合成
     *   基于统计参数的语音合成,优点是能适应的韵律特征的范围较宽,合成器比特率低,资源占用小,性能高,音质适中
     * 1 波形拼接合成
     *   基于高质量音库提取学习合成,资源占用相对较高,音质较好,更加贴近真实发音,但没有参数合成稳定
     *
     * @param method
     */
    public void setMethod(int method) {
        payload.put(TTSConstant.PROP_TTS_METHOD, method);
    }

    /**
     * 开始语音合成
     *
     * @throws Exception
     */
    @Override
    public void start() throws Exception {
        super.start();
        state = STATE_REQUEST_CONFIRMED;
        completeLatch = new CountDownLatch(1);
        listener.setCompleteLatch(completeLatch);
    }

    /**
     * 等待语音合成结束
     *
     * @param millsSeconds 最多等待多久就按超时处理
     * @throws Exception
     */
    public void waitForComplete(long millsSeconds) throws Exception {
        completeLatch.await(millsSeconds, TimeUnit.MILLISECONDS);
    }

    /**
     * 等待语音合成结束
     *
     * @throws Exception
     */
    public void waitForComplete() throws Exception {
        completeLatch.await();
    }

    /**
     * 关闭连接
     */
    public void close() {
        conn.close();
    }

    void markClosed() {
        state = STATE_CLOSED;
        if (completeLatch != null) {
            completeLatch.countDown();
        }
    }
}




© 2015 - 2024 Weber Informatics LLC | Privacy Policy