前段时间做了一个变声软件，思路是麦克风录入声音保存到本地，然后把声音转化成与之对应类型的声音。
做的时候在网上看了好多博客，也去做语音识别大厂官网去查阅了相关资料，然后利用这个插件VoiceChangerFilter简单的完成了一些基础功能，如果想要完全智能化，推荐还是购买对接（例如：科大讯飞等），官网上边都有试用版本也可以接入SDK尝试一下。

    //字节化成音频public AudioClip ConvertBytesToClip(byte[] _bs){//clip.length;float[] samples = new float[_bs.Length / 2];float rescaleFactor = 32767;short st = 0;float ft = 0;for (int i = 0; i < _bs.Length; i += 2){st = BitConverter.ToInt16(_bs, i);ft = st / rescaleFactor;samples[i / 2] = ft;}AudioClip cp = AudioClip.Create("clip", samples.Length, 1, audioFrequency, false);cp.SetData(samples, 0);return cp;}

  //开始录音IEnumerator InitStart(){yield return null;canRecord = true;source.clip = null;if (Microphone.devices.Length == 0){canRecord = false;//print("没有声音输入设备");Debug.LogWarning("没有声音输入设备");yield break;}sendt = 1.0f / SendRate;recording = Microphone.Start(null, false, recordLengh, audioFrequency);}

    //停止录音public void stopLuYin(){if (_IsStartRecord){_IsStartRecord = false;Microphone.End(null);List<byte> lss = new List<byte>();sendBuffer.ForEach((s) =>{lss.AddRange(s);print(s);});if (lss.Count <= 0) return;source.clip = ConvertBytesToClip(lss.ToArray());//source.clip = recording;SavWav.Save("test11", source.clip);//MediaCtr.OBJS["变声按钮"].SetActive(true);
#if UNITY_IOS// ... iOS项目编译
#elif UNITY_ANDROID// ... apk 或 iOS项目编译
#elif UNITY_EDITORAssetDatabase.Refresh();
#endifsendBuffer.Clear();VoiceChangerDemoScene.instance.sampleVoiceClip = source.clip;print("停止录音");//MediaCtr.instance.OverLuYin();}else{print("已经停止录音了");return;}MediaCtr.instance.OverLuYin();}

效果图：