mirror of
https://github.com/k2-fsa/sherpa-onnx.git
synced 2026-11-10 21:11:03 +08:00
Fangjun Kuang
708860f4fb
This pull request introduces a comprehensive C++ API for offline audio source separation. It enables developers to easily integrate Spleeter and UVR (MDX-Net) models into their C++ projects, providing functionalities to separate audio into distinct stems like vocals and accompaniment. The API is accompanied by practical examples demonstrating its usage and integration into the build system.
99 lines
3.1 KiB
C
99 lines
3.1 KiB
C
// c-api-examples/source-separation-spleeter-c-api.c
|
|
//
|
|
// Copyright (c) 2026 Xiaomi Corporation
|
|
|
|
// This file demonstrates how to use the source-separation C API
|
|
// with the Spleeter 2-stems model.
|
|
//
|
|
// Usage:
|
|
//
|
|
// 1. Download the test model and audio
|
|
//
|
|
// wget https://github.com/k2-fsa/sherpa-onnx/releases/download/source-separation-models/sherpa-onnx-spleeter-2stems-fp16.tar.bz2
|
|
// tar xjf sherpa-onnx-spleeter-2stems-fp16.tar.bz2
|
|
//
|
|
// wget https://github.com/k2-fsa/sherpa-onnx/releases/download/source-separation-models/qi-feng-le-zh.wav
|
|
//
|
|
// 2. Build
|
|
//
|
|
// cmake -DSHERPA_ONNX_ENABLE_C_API=ON ..
|
|
// make source-separation-spleeter-c-api
|
|
//
|
|
// 3. Run
|
|
//
|
|
// ./bin/source-separation-spleeter-c-api
|
|
|
|
#include <stdio.h>
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
|
|
#include "sherpa-onnx/c-api/c-api.h"
|
|
|
|
int32_t main() {
|
|
SherpaOnnxOfflineSourceSeparationConfig config;
|
|
memset(&config, 0, sizeof(config));
|
|
config.model.spleeter.vocals =
|
|
"./sherpa-onnx-spleeter-2stems-fp16/vocals.fp16.onnx";
|
|
config.model.spleeter.accompaniment =
|
|
"./sherpa-onnx-spleeter-2stems-fp16/accompaniment.fp16.onnx";
|
|
config.model.num_threads = 1;
|
|
|
|
const SherpaOnnxOfflineSourceSeparation *ss =
|
|
SherpaOnnxCreateOfflineSourceSeparation(&config);
|
|
if (!ss) {
|
|
fprintf(stderr, "Failed to create source separation engine\n");
|
|
return -1;
|
|
}
|
|
|
|
const SherpaOnnxMultiChannelWave *wave =
|
|
SherpaOnnxReadWaveMultiChannel("./qi-feng-le-zh.wav");
|
|
if (!wave) {
|
|
fprintf(stderr, "Failed to read ./qi-feng-le-zh.wav\n");
|
|
SherpaOnnxDestroyOfflineSourceSeparation(ss);
|
|
return -1;
|
|
}
|
|
|
|
fprintf(stdout, "Input wave: channels=%d, samples_per_channel=%d, sample_rate=%d\n",
|
|
wave->num_channels, wave->num_samples, wave->sample_rate);
|
|
|
|
int32_t num_channels = wave->num_channels;
|
|
|
|
const SherpaOnnxSourceSeparationOutput *output =
|
|
SherpaOnnxOfflineSourceSeparationProcess(
|
|
ss, wave->samples, num_channels, wave->num_samples,
|
|
wave->sample_rate);
|
|
|
|
if (!output) {
|
|
fprintf(stderr, "Source separation failed\n");
|
|
SherpaOnnxFreeMultiChannelWave(wave);
|
|
SherpaOnnxDestroyOfflineSourceSeparation(ss);
|
|
return -1;
|
|
}
|
|
|
|
fprintf(stdout, "Output: %d stems, sample_rate=%d\n", output->num_stems,
|
|
output->sample_rate);
|
|
|
|
// Write each stem to a separate multi-channel wave file.
|
|
const char *stem_names[] = {"vocals", "accompaniment"};
|
|
for (int32_t s = 0;
|
|
s < output->num_stems &&
|
|
s < (int32_t)(sizeof(stem_names) / sizeof(stem_names[0]));
|
|
++s) {
|
|
int32_t nc = output->stems[s].num_channels;
|
|
int32_t ns = output->stems[s].n;
|
|
char filename[256];
|
|
snprintf(filename, sizeof(filename), "%s.wav", stem_names[s]);
|
|
SherpaOnnxWriteWaveMultiChannel((const float *const *)output->stems[s].samples, ns,
|
|
output->sample_rate, nc, filename);
|
|
fprintf(stdout, "Saved %s (%d channels, %d samples, %d Hz)\n", filename,
|
|
nc, ns, output->sample_rate);
|
|
}
|
|
|
|
// Cleanup
|
|
SherpaOnnxDestroySourceSeparationOutput(output);
|
|
SherpaOnnxFreeMultiChannelWave(wave);
|
|
SherpaOnnxDestroyOfflineSourceSeparation(ss);
|
|
|
|
return 0;
|
|
}
|