使用C#实现方式文本转语音(TTS)及多音频合并实用指南

佚名 2026-08-28 12:40:01

平时做技术实践时，很多问题不是概念不会，而是细节没串起来。拿“采用C#实现方法文本转语音(TTS)及多音频合并”来说，它看着像小点，放到项目里常会牵出环境、配置、兼容性和维护成本。下面按实际采用顺序，把思路、关键写法和容易踩坑的地方讲清楚，便于大家直接对照操作。

一、文字转语音核心实现（System.Speech）

using System.Speech.Synthesis;
using System.Speech.AudioFormat;
public class VoiceSynthesizer
{
    private readonly SpeechSynthesizer _synthesizer;
    private readonly SpeechAudioFormatInfo _audioFormat;
    public VoiceSynthesizer()
    {
        _synthesizer = new SpeechSynthesizer();
        _synthesizer.SetOutputToNull(); // 禁用默认输出
        _audioFormat = new SpeechAudioFormatInfo(
            16000, // 采样率
            AudioBitsPerSample.Sixteen,
            AudioChannel.Mono);
    }
    /// <summary>
    /// 生成语音文件
    /// </summary>
    public string GenerateSpeech(string text, string outputPath)
    {
        using var stream = new MemoryStream();
        _synthesizer.SetOutputToWaveStream(stream);
        // 设置语音参数
        _synthesizer.Rate = 0;      // 语速（-10到10）
        _synthesizer.Volume = 100;  // 音量（0-100）
        _synthesizer.SelectVoiceByHints(VoiceGender.Female); // 选择语音
        _synthesizer.Speak(text);
        _synthesizer.SetOutputToNull();
        File.WriteAllBytes(outputPath, stream.ToArray());
        return outputPath;
    }
}

二、多音频合并实现（NAudio库）

using NAudio.Wave;

public class AudioMerger
{
    /// <summary>
    /// 合并多个WAV文件
    /// </summary>
    public void MergeAudioFiles(List<string> inputFiles, string outputFile)
    {
        using var output = new WaveFileWriter(outputFile, new WaveFormat(16000, 1));
        
        foreach (var file in inputFiles)
        {
            using var input = new WaveFileReader(file);
            input.CopyTo(output);
        }
    }

    /// <summary>
    /// 带淡入淡出的音频合并
    /// </summary>
    public void FadeMerge(List<string> files, string output, int fadeInMs = 500, int fadeOutMs = 500)
    {
        var mixer = new MixingSampleProvider(new[] { GetAudioProvider(files[0]) });
        
        for (int i = 1; i < files.Count; i++)
        {
            mixer.Add(input);
        }

        // 添加淡入淡出效果
        mixer = mixer.AppendFadeIn(fadeInMs).AppendFadeOut(fadeOutMs);
        
        WaveFileWriter.CreateWaveFile(output, mixer);
    }

    private ISampleProvider GetAudioProvider(string path)
    {
        var reader = new AudioFileReader(path);
        return reader.ToSampleProvider();
    }
}

三、完整工作流程示例

public class TtsWorkflow
{
    private readonly VoiceSynthesizer _tts;
    private readonly AudioMerger _merger;

    public TtsWorkflow()
    {
        _tts = new VoiceSynthesizer();
        _merger = new AudioMerger();
    }

    public void ProcessBatch(List<string> texts, string outputDir)
    {
        var tempFiles = new List<string>();

        try
        {
            // 分段生成语音
            foreach (var text in texts)
            {
                var tempFile = Path.Combine(outputDir, $"temp_{Guid.NewGuid()}.wav");
                _tts.GenerateSpeech(text, tempFile);
                tempFiles.Add(tempFile);
            }

            // 合并音频
            _merger.MergeAudioFiles(tempFiles, Path.Combine(outputDir, "output.wav"));
        }
        finally
        {
            // 清理临时文件
            foreach (var file in tempFiles)
            {
                File.Delete(file);
            }
        }
    }
}

四、关键功能扩展

1. 语音参数设置

// 设置语音属性
public void ConfigureVoice(VoiceGender gender, VoiceAge age, int rate = 0, int volume = 100)
{
    _synthesizer.SelectVoiceByHints(gender, age);
    _synthesizer.Rate = rate;
    _synthesizer.Volume = volume;
}

2. 异步处理优化

public async Task<string> GenerateSpeechAsync(string text, string outputPath)
{
    return await Task.Run(() => GenerateSpeech(text, outputPath));
}

3. SSML标记兼容

public string GenerateSsmlSpeech(string ssml)
{
    var promptBuilder = new PromptBuilder();
    promptBuilder.LoadSsml(ssml);
    using var stream = new MemoryStream();
    _synthesizer.SetOutputToWaveStream(stream);
    _synthesizer.Speak(promptBuilder);
    return File.ReadAllBytes(stream.ToArray()).ToBase64();
}

五、性能优化

1.语音缓存机制

private static readonly Dictionary<string, byte[]> _speechCache = new();

public string GetCachedSpeech(string text)
{
    if (!_speechCache.TryGetValue(text, out var data))
    {
        data = GenerateSpeech(text, Path.GetTempFileName());
        _speechCache[text] = data;
    }
    return Convert.ToBase64String(data);
}

2.批量处理优化

public void BatchProcess(List<TextSegment> segments)
{
    Parallel.ForEach(segments, segment => 
    {
        var tempFile = _tts.GenerateSpeech(segment.Text, Path.GetTempFileName());
        lock (_merger)
        {
            _merger.AppendToOutput(tempFile);
        }
    });
}

六、异常处理与日志

public class TtsExceptionHandler
{
    public void HandleException(Exception ex)
    {
        if (ex is InvalidOperationException)
        {
            Log($"语音引擎未初始化: {ex.InnerException?.Message}");
            InitializeEngine();
        }
        else if (ex is IOException)
        {
            Log($"文件写入失败: {ex.Message}");
            RetryOperation(() => File.WriteAllText(outputPath, content));
        }
        else
        {
            Log($"未知错误: {ex.StackTrace}");
        }
    }

    private void Log(string message)
    {
        File.AppendAllText("error.log", $"{DateTime.Now}: {message}n");
    }
}

七、部署与依赖管理

1.NuGet依赖

<PackageReference Include="System.Speech" Version="6.0.0" />
<PackageReference Include="NAudio" Version="2.1.0" />

2.运行环境要求

Windows 10/11（需安装语音引擎）
.NET 6.0或更高版本
至少2GB可用内存（处理长文本时）

参考代码 C# TTS语音朗读同时合成语音(文字转语音) www.youwenfan.com/contentcsp/116289.html

八、应用场景示例

1.有声读物生成

var texts = File.ReadAllLines("book.txt")
                .Select(line => line.Trim())
                .Where(line => !string.IsNullOrEmpty(line))
                .ToList();

var processor = new TtsWorkflow();
processor.ProcessBatch(texts, "audiobook_output");

2.实时语音播报

var synthesizer = new VoiceSynthesizer();
synthesizer.SetOutputToDefaultAudioDevice();
synthesizer.SpeakAsync("当前温度：25℃");

九、高级功能实现

1.情感语音合成

public void SetEmotion(VoiceEmotion emotion)
{
    var prompt = new PromptBuilder();
    prompt.AppendSsmlMarkup($"<prosody rate='{emotion.Rate}' pitch='{emotion.Pitch}'>");
    prompt.AppendText("需要强调的文本");
    prompt.AppendSsmlMarkup("</prosody>");
    _synthesizer.Speak(prompt);
}

public enum VoiceEmotion
{
    Neutral,
    Happy,
    Angry,
    Sad
}

2.多语言兼容

public void SwitchLanguage(string cultureCode)
{
    var voices = _synthesizer.GetInstalledVoices();
    var targetVoice = voices.FirstOrDefault(v => 
        v.VoiceInfo.Culture.Name.Equals(cultureCode, StringComparison.OrdinalIgnoreCase));
    
    if (targetVoice != null)
    {
        _synthesizer.SelectVoice(targetVoice.VoiceInfo.Name);
    }
}