将音频从ByteBuffer（从exoplayer）转换为float *（IPLFloat32），然后再返回以进行处理

Question

我正在为ExoPlayer（使用kotlin / Java）编写一个自定义AudioProcessor，它将传入的音频数据，将其传递到Steam Audio API（在C ++中），使用iplApplyBinauralEffect进行处理并将其传递回ExoPlayer 。当ExoPlayer从queueInput提供的ByteBuffer需要转换为Steam期望的float *（IPLfloat32）时，就会出现问题。如何将ByteBuffer转换为Steam可以处理的float *？

ByteBuffer表示它是小端的，所以我转换为大端的，但是它仍然返回大量且很小的值。

The docs for Steam Audio API are here
and the docs for ExoPlayer are here

音频缓冲区IPLAudioBuffer is described here

（从AudioProcessor子类传递）

    public ByteBuffer queueInput(ByteBuffer inputBuffer, ByteBuffer buffer, int position, int length) {
        //inputBuffer.order(ByteOrder.BIG_ENDIAN);
        FloatBuffer floatBuffer = inputBuffer.order(ByteOrder.BIG_ENDIAN).asFloatBuffer();
        float[] input = new float[floatBuffer.limit()];
        floatBuffer.get(input);
        // get rendered output
        float[] convertedArray = queueInput(playerIndex, input, floatBuffer.limit());
        buffer = ByteBuffer.allocate(inputBuffer.limit() * 2);
        FloatBuffer outputFloatBuffer = buffer.asFloatBuffer();
        outputFloatBuffer.put(convertedArray);
        buffer.rewind();
        return buffer;
    }

（通过jni）

    queueInput(JNIEnv *env, jobject instance, jint playerIndex, jfloatArray bytes, jint len) {
    jfloat *input = env->GetFloatArrayElements(bytes, nullptr);
    int inputaudio = engine->getOutput(static_cast<unsigned int>(playerIndex), input, len, input);
    jfloatArray result;
    result = env->NewFloatArray(inputaudio);
    env->SetFloatArrayRegion(result, 0, inputaudio, input);

    free(input);
    env->ReleaseFloatArrayElements(bytes, input, JNI_COMMIT);
    return result;
    }

（传递给SteamPlayer）

    int SteamPlayer::getOutput(float *input, int len, float *output) {
        std::vector<float> outputaudioframe(static_cast<unsigned int>(2 * framesize));
        std::vector<float> outputaudio;

        int numFrames = len / framesize;

        IPLAudioBuffer inbuffer{ stereo, framesize, input };
        inbuffer.format.numSpeakers = 2;
        inbuffer.format.speakerDirections = &inDirections;

        IPLAudioBuffer outbuffer{stereo, framesize, outputaudioframe.data()};
        outbuffer.format.numSpeakers = 2;
        outDirections.x = x;
        outDirections.y = y;
        outDirections.z = z;
        outbuffer.format.speakerDirections = &outDirections;

        for (auto i = 0; i < numFrames; ++i) {
            iplApplyBinauralEffect(effect, renderer, inbuffer, outDirections, IPL_HRTFINTERPOLATION_NEAREST, outbuffer);
            std::copy(std::begin(outputaudioframe), std::end(outputaudioframe), std::back_inserter(outputaudio));
            inbuffer.interleavedBuffer += framesize;
        }
        for (int i = 0; i < outputaudio.size(); ++i) {
            outputaudio[i] *= 32767;
        }
        std::memcpy(output, outputaudio.data(), outputaudio.size());
        return outputaudio.size();

    }

这是“ convertedArray”行之前的一些传入数据的屏幕截图

任何帮助将不胜感激。