services/audioflinger/test-resample.cpp

/*
 * Copyright (C) 2012 The Android Open Source Project
 *
 * 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.
 */

#include "AudioResampler.h"
#include <media/AudioBufferProvider.h>
#include <unistd.h>
#include <stdio.h>
#include <stdlib.h>
#include <fcntl.h>
#include <string.h>
#include <sys/mman.h>
#include <sys/stat.h>
#include <errno.h>
#include <inttypes.h>
#include <time.h>
#include <math.h>
#include <audio_utils/primitives.h>
#include <audio_utils/sndfile.h>
#include <utils/Vector.h>

using namespace android;

static bool gVerbose = false;

static int usage(const char* name) {
    fprintf(stderr,"Usage: %s [-p] [-f] [-F] [-v] [-c channels]"
                   " [-q {dq|lq|mq|hq|vhq|dlq|dmq|dhq}]"
                   " [-i input-sample-rate] [-o output-sample-rate]"
                   " [-O csv] [-P csv] [<input-file>]"
                   " <output-file>\n", name);
    fprintf(stderr,"    -p    enable profiling\n");
    fprintf(stderr,"    -f    enable filter profiling\n");
    fprintf(stderr,"    -F    enable floating point -q {dlq|dmq|dhq} only");
    fprintf(stderr,"    -v    verbose : log buffer provider calls\n");
    fprintf(stderr,"    -c    # channels (1-2 for lq|mq|hq; 1-8 for dlq|dmq|dhq)\n");
    fprintf(stderr,"    -q    resampler quality\n");
    fprintf(stderr,"              dq  : default quality\n");
    fprintf(stderr,"              lq  : low quality\n");
    fprintf(stderr,"              mq  : medium quality\n");
    fprintf(stderr,"              hq  : high quality\n");
    fprintf(stderr,"              vhq : very high quality\n");
    fprintf(stderr,"              dlq : dynamic low quality\n");
    fprintf(stderr,"              dmq : dynamic medium quality\n");
    fprintf(stderr,"              dhq : dynamic high quality\n");
    fprintf(stderr,"    -i    input file sample rate (ignored if input file is specified)\n");
    fprintf(stderr,"    -o    output file sample rate\n");
    fprintf(stderr,"    -O    # frames output per call to resample() in CSV format\n");
    fprintf(stderr,"    -P    # frames provided per call to resample() in CSV format\n");
    return -1;
}

// Convert a list of integers in CSV format to a Vector of those values.
// Returns the number of elements in the list, or -1 on error.
int parseCSV(const char *string, Vector<int>& values)
{
    // pass 1: count the number of values and do syntax check
    size_t numValues = 0;
    bool hadDigit = false;
    for (const char *p = string; ; ) {
        switch (*p++) {
        case '0': case '1': case '2': case '3': case '4':
        case '5': case '6': case '7': case '8': case '9':
            hadDigit = true;
            break;
        case '\0':
            if (hadDigit) {
                // pass 2: allocate and initialize vector of values
                values.resize(++numValues);
                values.editItemAt(0) = atoi(p = optarg);
                for (size_t i = 1; i < numValues; ) {
                    if (*p++ == ',') {
                        values.editItemAt(i++) = atoi(p);
                    }
                }
                return numValues;
            }
            // fall through
        case ',':
            if (hadDigit) {
                hadDigit = false;
                numValues++;
                break;
            }
            // fall through
        default:
            return -1;
        }
    }
}

int main(int argc, char* argv[]) {
    const char* const progname = argv[0];
    bool profileResample = false;
    bool profileFilter = false;
    bool useFloat = false;
    int channels = 1;
    int input_freq = 0;
    int output_freq = 0;
    AudioResampler::src_quality quality = AudioResampler::DEFAULT_QUALITY;
    Vector<int> Ovalues;
    Vector<int> Pvalues;

    int ch;
    while ((ch = getopt(argc, argv, "pfFvc:q:i:o:O:P:")) != -1) {
        switch (ch) {
        case 'p':
            profileResample = true;
            break;
        case 'f':
            profileFilter = true;
            break;
        case 'F':
            useFloat = true;
            break;
        case 'v':
            gVerbose = true;
            break;
        case 'c':
            channels = atoi(optarg);
            break;
        case 'q':
            if (!strcmp(optarg, "dq"))
                quality = AudioResampler::DEFAULT_QUALITY;
            else if (!strcmp(optarg, "lq"))
                quality = AudioResampler::LOW_QUALITY;
            else if (!strcmp(optarg, "mq"))
                quality = AudioResampler::MED_QUALITY;
            else if (!strcmp(optarg, "hq"))
                quality = AudioResampler::HIGH_QUALITY;
            else if (!strcmp(optarg, "vhq"))
                quality = AudioResampler::VERY_HIGH_QUALITY;
            else if (!strcmp(optarg, "dlq"))
                quality = AudioResampler::DYN_LOW_QUALITY;
            else if (!strcmp(optarg, "dmq"))
                quality = AudioResampler::DYN_MED_QUALITY;
            else if (!strcmp(optarg, "dhq"))
                quality = AudioResampler::DYN_HIGH_QUALITY;
            else {
                usage(progname);
                return -1;
            }
            break;
        case 'i':
            input_freq = atoi(optarg);
            break;
        case 'o':
            output_freq = atoi(optarg);
            break;
        case 'O':
            if (parseCSV(optarg, Ovalues) < 0) {
                fprintf(stderr, "incorrect syntax for -O option\n");
                return -1;
            }
            break;
        case 'P':
            if (parseCSV(optarg, Pvalues) < 0) {
                fprintf(stderr, "incorrect syntax for -P option\n");
                return -1;
            }
            break;
        case '?':
        default:
            usage(progname);
            return -1;
        }
    }

    if (channels < 1
            || channels > (quality < AudioResampler::DYN_LOW_QUALITY ? 2 : 8)) {
        fprintf(stderr, "invalid number of audio channels %d\n", channels);
        return -1;
    }
    if (useFloat && quality < AudioResampler::DYN_LOW_QUALITY) {
        fprintf(stderr, "float processing is only possible for dynamic resamplers\n");
        return -1;
    }

    argc -= optind;
    argv += optind;

    const char* file_in = NULL;
    const char* file_out = NULL;
    if (argc == 1) {
        file_out = argv[0];
    } else if (argc == 2) {
        file_in = argv[0];
        file_out = argv[1];
    } else {
        usage(progname);
        return -1;
    }

    // ----------------------------------------------------------

    size_t input_size;
    void* input_vaddr;
    if (argc == 2) {
        SF_INFO info;
        info.format = 0;
        SNDFILE *sf = sf_open(file_in, SFM_READ, &info);
        if (sf == NULL) {
            perror(file_in);
            return EXIT_FAILURE;
        }
        input_size = info.frames * info.channels * sizeof(short);
        input_vaddr = malloc(input_size);
        (void) sf_readf_short(sf, (short *) input_vaddr, info.frames);
        sf_close(sf);
        channels = info.channels;
        input_freq = info.samplerate;
    } else {
        // data for testing is exactly (input sampling rate/1000)/2 seconds
        // so 44.1khz input is 22.05 seconds
        double k = 1000; // Hz / s
        double time = (input_freq / 2) / k;
        size_t input_frames = size_t(input_freq * time);
        input_size = channels * sizeof(int16_t) * input_frames;
        input_vaddr = malloc(input_size);
        int16_t* in = (int16_t*)input_vaddr;
        for (size_t i=0 ; i<input_frames ; i++) {
            double t = double(i) / input_freq;
            double y = sin(M_PI * k * t * t);
            int16_t yi = floor(y * 32767.0 + 0.5);
            for (int j = 0; j < channels; j++) {
                in[i*channels + j] = yi / (1 + j);
            }
        }
    }
    size_t input_framesize = channels * sizeof(int16_t);
    size_t input_frames = input_size / input_framesize;

    // For float processing, convert input int16_t to float array
    if (useFloat) {
        void *new_vaddr;

        input_framesize = channels * sizeof(float);
        input_size = input_frames * input_framesize;
        new_vaddr = malloc(input_size);
        memcpy_to_float_from_i16(reinterpret_cast<float*>(new_vaddr),
                reinterpret_cast<int16_t*>(input_vaddr), input_frames * channels);
        free(input_vaddr);
        input_vaddr = new_vaddr;
    }

    // ----------------------------------------------------------

    class Provider: public AudioBufferProvider {
        const void*     mAddr;      // base address
        const size_t    mNumFrames; // total frames
        const size_t    mFrameSize; // size of each frame in bytes
        size_t          mNextFrame; // index of next frame to provide
        size_t          mUnrel;     // number of frames not yet released
        const Vector<int> mPvalues; // number of frames provided per call
        size_t          mNextPidx;  // index of next entry in mPvalues to use
    public:
        Provider(const void* addr, size_t frames, size_t frameSize, const Vector<int>& Pvalues)
          : mAddr(addr),
            mNumFrames(frames),
            mFrameSize(frameSize),
            mNextFrame(0), mUnrel(0), mPvalues(Pvalues), mNextPidx(0) {
        }
        virtual status_t getNextBuffer(Buffer* buffer,
                int64_t pts = kInvalidPTS) {
            (void)pts; // suppress warning
            size_t requestedFrames = buffer->frameCount;
            if (requestedFrames > mNumFrames - mNextFrame) {
                buffer->frameCount = mNumFrames - mNextFrame;
            }
            if (!mPvalues.isEmpty()) {
                size_t provided = mPvalues[mNextPidx++];
                printf("mPvalue[%zu]=%zu not %zu\n", mNextPidx-1, provided, buffer->frameCount);
                if (provided < buffer->frameCount) {
                    buffer->frameCount = provided;
                }
                if (mNextPidx >= mPvalues.size()) {
                    mNextPidx = 0;
                }
            }
            if (gVerbose) {
                printf("getNextBuffer() requested %zu frames out of %zu frames available,"
                        " and returned %zu frames\n",
                        requestedFrames, (size_t) (mNumFrames - mNextFrame), buffer->frameCount);
            }
            mUnrel = buffer->frameCount;
            if (buffer->frameCount > 0) {
                buffer->raw = (char *)mAddr + mFrameSize * mNextFrame;
                return NO_ERROR;
            } else {
                buffer->raw = NULL;
                return NOT_ENOUGH_DATA;
            }
        }
        virtual void releaseBuffer(Buffer* buffer) {
            if (buffer->frameCount > mUnrel) {
                fprintf(stderr, "ERROR releaseBuffer() released %zu frames but only %zu available "
                        "to release\n", buffer->frameCount, mUnrel);
                mNextFrame += mUnrel;
                mUnrel = 0;
            } else {
                if (gVerbose) {
                    printf("releaseBuffer() released %zu frames out of %zu frames available "
                            "to release\n", buffer->frameCount, mUnrel);
                }
                mNextFrame += buffer->frameCount;
                mUnrel -= buffer->frameCount;
            }
            buffer->frameCount = 0;
            buffer->raw = NULL;
        }
        void reset() {
            mNextFrame = 0;
        }
    } provider(input_vaddr, input_frames, input_framesize, Pvalues);

    if (gVerbose) {
        printf("%zu input frames\n", input_frames);
    }

    int bit_depth = useFloat ? 32 : 16;
    int output_channels = channels > 2 ? channels : 2; // output is at least stereo samples
    size_t output_framesize = output_channels * (useFloat ? sizeof(float) : sizeof(int32_t));
    size_t output_frames = ((int64_t) input_frames * output_freq) / input_freq;
    size_t output_size = output_frames * output_framesize;

    if (profileFilter) {
        // Check how fast sample rate changes are that require filter changes.
        // The delta sample rate changes must indicate a downsampling ratio,
        // and must be larger than 10% changes.
        //
        // On fast devices, filters should be generated between 0.1ms - 1ms.
        // (single threaded).
        AudioResampler* resampler = AudioResampler::create(bit_depth, channels,
                8000, quality);
        int looplimit = 100;
        timespec start, end;
        clock_gettime(CLOCK_MONOTONIC, &start);
        for (int i = 0; i < looplimit; ++i) {
            resampler->setSampleRate(9000);
            resampler->setSampleRate(12000);
            resampler->setSampleRate(20000);
            resampler->setSampleRate(30000);
        }
        clock_gettime(CLOCK_MONOTONIC, &end);
        int64_t start_ns = start.tv_sec * 1000000000LL + start.tv_nsec;
        int64_t end_ns = end.tv_sec * 1000000000LL + end.tv_nsec;
        int64_t time = end_ns - start_ns;
        printf("%.2f sample rate changes with filter calculation/sec\n",
                looplimit * 4 / (time / 1e9));

        // Check how fast sample rate changes are without filter changes.
        // This should be very fast, probably 0.1us - 1us per sample rate
        // change.
        resampler->setSampleRate(1000);
        looplimit = 1000;
        clock_gettime(CLOCK_MONOTONIC, &start);
        for (int i = 0; i < looplimit; ++i) {
            resampler->setSampleRate(1000+i);
        }
        clock_gettime(CLOCK_MONOTONIC, &end);
        start_ns = start.tv_sec * 1000000000LL + start.tv_nsec;
        end_ns = end.tv_sec * 1000000000LL + end.tv_nsec;
        time = end_ns - start_ns;
        printf("%.2f sample rate changes without filter calculation/sec\n",
                looplimit / (time / 1e9));
        resampler->reset();
        delete resampler;
    }

    void* output_vaddr = malloc(output_size);
    AudioResampler* resampler = AudioResampler::create(bit_depth, channels,
            output_freq, quality);


    /* set volume precision to 12 bits, so the volume scale is 1<<12.
     * The output int32_t is represented as Q4.27, with 4 bits of guard
     * followed by the int16_t Q.15 portion, and then 12 trailing bits of
     * additional precision.
     *
     * Generally 0 < volumePrecision <= 14 (due to the limits of
     * int16_t values for Volume). volumePrecision cannot be 0 due
     * to rounding and shifts.
     */
    const int volumePrecision = 12; // in bits

    resampler->setSampleRate(input_freq);
    resampler->setVolume(1 << volumePrecision, 1 << volumePrecision);

    if (profileResample) {
        /*
         * For profiling on mobile devices, upon experimentation
         * it is better to run a few trials with a shorter loop limit,
         * and take the minimum time.
         *
         * Long tests can cause CPU temperature to build up and thermal throttling
         * to reduce CPU frequency.
         *
         * For frequency checks (index=0, or 1, etc.):
         * "cat /sys/devices/system/cpu/cpu${index}/cpufreq/scaling_*_freq"
         *
         * For temperature checks (index=0, or 1, etc.):
         * "cat /sys/class/thermal/thermal_zone${index}/temp"
         *
         * Another way to avoid thermal throttling is to fix the CPU frequency
         * at a lower level which prevents excessive temperatures.
         */
        const int trials = 4;
        const int looplimit = 4;
        timespec start, end;
        int64_t time = 0;

        for (int n = 0; n < trials; ++n) {
            clock_gettime(CLOCK_MONOTONIC, &start);
            for (int i = 0; i < looplimit; ++i) {
                resampler->resample((int*) output_vaddr, output_frames, &provider);
                provider.reset(); //  during benchmarking reset only the provider
            }
            clock_gettime(CLOCK_MONOTONIC, &end);
            int64_t start_ns = start.tv_sec * 1000000000LL + start.tv_nsec;
            int64_t end_ns = end.tv_sec * 1000000000LL + end.tv_nsec;
            int64_t diff_ns = end_ns - start_ns;
            if (n == 0 || diff_ns < time) {
                time = diff_ns;   // save the best out of our trials.
            }
        }
        // Mfrms/s is "Millions of output frames per second".
        printf("quality: %d  channels: %d  msec: %" PRId64 "  Mfrms/s: %.2lf\n",
                quality, channels, time/1000000, output_frames * looplimit / (time / 1e9) / 1e6);
        resampler->reset();
    }

    memset(output_vaddr, 0, output_size);
    if (gVerbose) {
        printf("resample() %zu output frames\n", output_frames);
    }
    if (Ovalues.isEmpty()) {
        Ovalues.push(output_frames);
    }
    for (size_t i = 0, j = 0; i < output_frames; ) {
        size_t thisFrames = Ovalues[j++];
        if (j >= Ovalues.size()) {
            j = 0;
        }
        if (thisFrames == 0 || thisFrames > output_frames - i) {
            thisFrames = output_frames - i;
        }
        resampler->resample((int*) output_vaddr + output_channels*i, thisFrames, &provider);
        i += thisFrames;
    }
    if (gVerbose) {
        printf("resample() complete\n");
    }
    resampler->reset();
    if (gVerbose) {
        printf("reset() complete\n");
    }
    delete resampler;
    resampler = NULL;

    // For float processing, convert output format from float to Q4.27,
    // which is then converted to int16_t for final storage.
    if (useFloat) {
        memcpy_to_q4_27_from_float(reinterpret_cast<int32_t*>(output_vaddr),
                reinterpret_cast<float*>(output_vaddr), output_frames * output_channels);
    }

    // mono takes left channel only (out of stereo output pair)
    // stereo and multichannel preserve all channels.
    int32_t* out = (int32_t*) output_vaddr;
    int16_t* convert = (int16_t*) malloc(output_frames * channels * sizeof(int16_t));

    // round to half towards zero and saturate at int16 (non-dithered)
    const int roundVal = (1<<(volumePrecision-1)) - 1; // volumePrecision > 0

    for (size_t i = 0; i < output_frames; i++) {
        for (int j = 0; j < channels; j++) {
            int32_t s = out[i * output_channels + j] + roundVal; // add offset here
            if (s < 0) {
                s = (s + 1) >> volumePrecision; // round to 0
                if (s < -32768) {
                    s = -32768;
                }
            } else {
                s = s >> volumePrecision;
                if (s > 32767) {
                    s = 32767;
                }
            }
            convert[i * channels + j] = int16_t(s);
        }
    }

    // write output to disk
    SF_INFO info;
    info.frames = 0;
    info.samplerate = output_freq;
    info.channels = channels;
    info.format = SF_FORMAT_WAV | SF_FORMAT_PCM_16;
    SNDFILE *sf = sf_open(file_out, SFM_WRITE, &info);
    if (sf == NULL) {
        perror(file_out);
        return EXIT_FAILURE;
    }
    (void) sf_writef_short(sf, convert, output_frames);
    sf_close(sf);

    return EXIT_SUCCESS;
}