hax/seek-tune
1
0
Fork 0
mirror of https://github.com/cgzirim/seek-tune.git synced 2025-12-18 09:24:19 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge pull request #28 from cgzirim/development

Browse source 
Development
This commit is contained in:
Chigozirim Igweamaka 2025-03-11 19:55:26 +01:00 • committed by GitHub
commit 900c8152d2
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
11 changed files with 734 additions and 207 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1
client/public/index.html
View file

@ -27,6 +27,7 @@
<title>SeekTune</title> <title>SeekTune</title>
</head> </head>
<body> <body>
<script src="/wasm_exec.js"></script>
<noscript>You need to enable JavaScript to run this app.</noscript> <noscript>You need to enable JavaScript to run this app.</noscript>
<div id="root"></div> <div id="root"></div>
<!-- <!--

BIN
client/public/main.wasm Executable file
View file

Binary file not shown.

561
client/public/wasm_exec.js Normal file
View file

@ -0,0 +1,561 @@
// Copyright 2018 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
"use strict";
(() => {
const enosys = () => {
const err = new Error("not implemented");
err.code = "ENOSYS";
return err;
};
if (!globalThis.fs) {
let outputBuf = "";
globalThis.fs = {
constants: { O_WRONLY: -1, O_RDWR: -1, O_CREAT: -1, O_TRUNC: -1, O_APPEND: -1, O_EXCL: -1 }, // unused
writeSync(fd, buf) {
outputBuf += decoder.decode(buf);
const nl = outputBuf.lastIndexOf("\n");
if (nl != -1) {
console.log(outputBuf.substring(0, nl));
outputBuf = outputBuf.substring(nl + 1);
}
return buf.length;
},
write(fd, buf, offset, length, position, callback) {
if (offset !== 0 || length !== buf.length || position !== null) {
callback(enosys());
return;
}
const n = this.writeSync(fd, buf);
callback(null, n);
},
chmod(path, mode, callback) { callback(enosys()); },
chown(path, uid, gid, callback) { callback(enosys()); },
close(fd, callback) { callback(enosys()); },
fchmod(fd, mode, callback) { callback(enosys()); },
fchown(fd, uid, gid, callback) { callback(enosys()); },
fstat(fd, callback) { callback(enosys()); },
fsync(fd, callback) { callback(null); },
ftruncate(fd, length, callback) { callback(enosys()); },
lchown(path, uid, gid, callback) { callback(enosys()); },
link(path, link, callback) { callback(enosys()); },
lstat(path, callback) { callback(enosys()); },
mkdir(path, perm, callback) { callback(enosys()); },
open(path, flags, mode, callback) { callback(enosys()); },
read(fd, buffer, offset, length, position, callback) { callback(enosys()); },
readdir(path, callback) { callback(enosys()); },
readlink(path, callback) { callback(enosys()); },
rename(from, to, callback) { callback(enosys()); },
rmdir(path, callback) { callback(enosys()); },
stat(path, callback) { callback(enosys()); },
symlink(path, link, callback) { callback(enosys()); },
truncate(path, length, callback) { callback(enosys()); },
unlink(path, callback) { callback(enosys()); },
utimes(path, atime, mtime, callback) { callback(enosys()); },
};
}
if (!globalThis.process) {
globalThis.process = {
getuid() { return -1; },
getgid() { return -1; },
geteuid() { return -1; },
getegid() { return -1; },
getgroups() { throw enosys(); },
pid: -1,
ppid: -1,
umask() { throw enosys(); },
cwd() { throw enosys(); },
chdir() { throw enosys(); },
}
}
if (!globalThis.crypto) {
throw new Error("globalThis.crypto is not available, polyfill required (crypto.getRandomValues only)");
}
if (!globalThis.performance) {
throw new Error("globalThis.performance is not available, polyfill required (performance.now only)");
}
if (!globalThis.TextEncoder) {
throw new Error("globalThis.TextEncoder is not available, polyfill required");
}
if (!globalThis.TextDecoder) {
throw new Error("globalThis.TextDecoder is not available, polyfill required");
}
const encoder = new TextEncoder("utf-8");
const decoder = new TextDecoder("utf-8");
globalThis.Go = class {
constructor() {
this.argv = ["js"];
this.env = {};
this.exit = (code) => {
if (code !== 0) {
console.warn("exit code:", code);
}
};
this._exitPromise = new Promise((resolve) => {
this._resolveExitPromise = resolve;
});
this._pendingEvent = null;
this._scheduledTimeouts = new Map();
this._nextCallbackTimeoutID = 1;
const setInt64 = (addr, v) => {
this.mem.setUint32(addr + 0, v, true);
this.mem.setUint32(addr + 4, Math.floor(v / 4294967296), true);
}
const setInt32 = (addr, v) => {
this.mem.setUint32(addr + 0, v, true);
}
const getInt64 = (addr) => {
const low = this.mem.getUint32(addr + 0, true);
const high = this.mem.getInt32(addr + 4, true);
return low + high * 4294967296;
}
const loadValue = (addr) => {
const f = this.mem.getFloat64(addr, true);
if (f === 0) {
return undefined;
}
if (!isNaN(f)) {
return f;
}
const id = this.mem.getUint32(addr, true);
return this._values[id];
}
const storeValue = (addr, v) => {
const nanHead = 0x7FF80000;
if (typeof v === "number" && v !== 0) {
if (isNaN(v)) {
this.mem.setUint32(addr + 4, nanHead, true);
this.mem.setUint32(addr, 0, true);
return;
}
this.mem.setFloat64(addr, v, true);
return;
}
if (v === undefined) {
this.mem.setFloat64(addr, 0, true);
return;
}
let id = this._ids.get(v);
if (id === undefined) {
id = this._idPool.pop();
if (id === undefined) {
id = this._values.length;
}
this._values[id] = v;
this._goRefCounts[id] = 0;
this._ids.set(v, id);
}
this._goRefCounts[id]++;
let typeFlag = 0;
switch (typeof v) {
case "object":
if (v !== null) {
typeFlag = 1;
}
break;
case "string":
typeFlag = 2;
break;
case "symbol":
typeFlag = 3;
break;
case "function":
typeFlag = 4;
break;
}
this.mem.setUint32(addr + 4, nanHead | typeFlag, true);
this.mem.setUint32(addr, id, true);
}
const loadSlice = (addr) => {
const array = getInt64(addr + 0);
const len = getInt64(addr + 8);
return new Uint8Array(this._inst.exports.mem.buffer, array, len);
}
const loadSliceOfValues = (addr) => {
const array = getInt64(addr + 0);
const len = getInt64(addr + 8);
const a = new Array(len);
for (let i = 0; i < len; i++) {
a[i] = loadValue(array + i * 8);
}
return a;
}
const loadString = (addr) => {
const saddr = getInt64(addr + 0);
const len = getInt64(addr + 8);
return decoder.decode(new DataView(this._inst.exports.mem.buffer, saddr, len));
}
const timeOrigin = Date.now() - performance.now();
this.importObject = {
_gotest: {
add: (a, b) => a + b,
},
gojs: {
// Go's SP does not change as long as no Go code is running. Some operations (e.g. calls, getters and setters)
// may synchronously trigger a Go event handler. This makes Go code get executed in the middle of the imported
// function. A goroutine can switch to a new stack if the current stack is too small (see morestack function).
// This changes the SP, thus we have to update the SP used by the imported function.
// func wasmExit(code int32)
"runtime.wasmExit": (sp) => {
sp >>>= 0;
const code = this.mem.getInt32(sp + 8, true);
this.exited = true;
delete this._inst;
delete this._values;
delete this._goRefCounts;
delete this._ids;
delete this._idPool;
this.exit(code);
},
// func wasmWrite(fd uintptr, p unsafe.Pointer, n int32)
"runtime.wasmWrite": (sp) => {
sp >>>= 0;
const fd = getInt64(sp + 8);
const p = getInt64(sp + 16);
const n = this.mem.getInt32(sp + 24, true);
fs.writeSync(fd, new Uint8Array(this._inst.exports.mem.buffer, p, n));
},
// func resetMemoryDataView()
"runtime.resetMemoryDataView": (sp) => {
sp >>>= 0;
this.mem = new DataView(this._inst.exports.mem.buffer);
},
// func nanotime1() int64
"runtime.nanotime1": (sp) => {
sp >>>= 0;
setInt64(sp + 8, (timeOrigin + performance.now()) * 1000000);
},
// func walltime() (sec int64, nsec int32)
"runtime.walltime": (sp) => {
sp >>>= 0;
const msec = (new Date).getTime();
setInt64(sp + 8, msec / 1000);
this.mem.setInt32(sp + 16, (msec % 1000) * 1000000, true);
},
// func scheduleTimeoutEvent(delay int64) int32
"runtime.scheduleTimeoutEvent": (sp) => {
sp >>>= 0;
const id = this._nextCallbackTimeoutID;
this._nextCallbackTimeoutID++;
this._scheduledTimeouts.set(id, setTimeout(
() => {
this._resume();
while (this._scheduledTimeouts.has(id)) {
// for some reason Go failed to register the timeout event, log and try again
// (temporary workaround for https://github.com/golang/go/issues/28975)
console.warn("scheduleTimeoutEvent: missed timeout event");
this._resume();
}
},
getInt64(sp + 8),
));
this.mem.setInt32(sp + 16, id, true);
},
// func clearTimeoutEvent(id int32)
"runtime.clearTimeoutEvent": (sp) => {
sp >>>= 0;
const id = this.mem.getInt32(sp + 8, true);
clearTimeout(this._scheduledTimeouts.get(id));
this._scheduledTimeouts.delete(id);
},
// func getRandomData(r []byte)
"runtime.getRandomData": (sp) => {
sp >>>= 0;
crypto.getRandomValues(loadSlice(sp + 8));
},
// func finalizeRef(v ref)
"syscall/js.finalizeRef": (sp) => {
sp >>>= 0;
const id = this.mem.getUint32(sp + 8, true);
this._goRefCounts[id]--;
if (this._goRefCounts[id] === 0) {
const v = this._values[id];
this._values[id] = null;
this._ids.delete(v);
this._idPool.push(id);
}
},
// func stringVal(value string) ref
"syscall/js.stringVal": (sp) => {
sp >>>= 0;
storeValue(sp + 24, loadString(sp + 8));
},
// func valueGet(v ref, p string) ref
"syscall/js.valueGet": (sp) => {
sp >>>= 0;
const result = Reflect.get(loadValue(sp + 8), loadString(sp + 16));
sp = this._inst.exports.getsp() >>> 0; // see comment above
storeValue(sp + 32, result);
},
// func valueSet(v ref, p string, x ref)
"syscall/js.valueSet": (sp) => {
sp >>>= 0;
Reflect.set(loadValue(sp + 8), loadString(sp + 16), loadValue(sp + 32));
},
// func valueDelete(v ref, p string)
"syscall/js.valueDelete": (sp) => {
sp >>>= 0;
Reflect.deleteProperty(loadValue(sp + 8), loadString(sp + 16));
},
// func valueIndex(v ref, i int) ref
"syscall/js.valueIndex": (sp) => {
sp >>>= 0;
storeValue(sp + 24, Reflect.get(loadValue(sp + 8), getInt64(sp + 16)));
},
// valueSetIndex(v ref, i int, x ref)
"syscall/js.valueSetIndex": (sp) => {
sp >>>= 0;
Reflect.set(loadValue(sp + 8), getInt64(sp + 16), loadValue(sp + 24));
},
// func valueCall(v ref, m string, args []ref) (ref, bool)
"syscall/js.valueCall": (sp) => {
sp >>>= 0;
try {
const v = loadValue(sp + 8);
const m = Reflect.get(v, loadString(sp + 16));
const args = loadSliceOfValues(sp + 32);
const result = Reflect.apply(m, v, args);
sp = this._inst.exports.getsp() >>> 0; // see comment above
storeValue(sp + 56, result);
this.mem.setUint8(sp + 64, 1);
} catch (err) {
sp = this._inst.exports.getsp() >>> 0; // see comment above
storeValue(sp + 56, err);
this.mem.setUint8(sp + 64, 0);
}
},
// func valueInvoke(v ref, args []ref) (ref, bool)
"syscall/js.valueInvoke": (sp) => {
sp >>>= 0;
try {
const v = loadValue(sp + 8);
const args = loadSliceOfValues(sp + 16);
const result = Reflect.apply(v, undefined, args);
sp = this._inst.exports.getsp() >>> 0; // see comment above
storeValue(sp + 40, result);
this.mem.setUint8(sp + 48, 1);
} catch (err) {
sp = this._inst.exports.getsp() >>> 0; // see comment above
storeValue(sp + 40, err);
this.mem.setUint8(sp + 48, 0);
}
},
// func valueNew(v ref, args []ref) (ref, bool)
"syscall/js.valueNew": (sp) => {
sp >>>= 0;
try {
const v = loadValue(sp + 8);
const args = loadSliceOfValues(sp + 16);
const result = Reflect.construct(v, args);
sp = this._inst.exports.getsp() >>> 0; // see comment above
storeValue(sp + 40, result);
this.mem.setUint8(sp + 48, 1);
} catch (err) {
sp = this._inst.exports.getsp() >>> 0; // see comment above
storeValue(sp + 40, err);
this.mem.setUint8(sp + 48, 0);
}
},
// func valueLength(v ref) int
"syscall/js.valueLength": (sp) => {
sp >>>= 0;
setInt64(sp + 16, parseInt(loadValue(sp + 8).length));
},
// valuePrepareString(v ref) (ref, int)
"syscall/js.valuePrepareString": (sp) => {
sp >>>= 0;
const str = encoder.encode(String(loadValue(sp + 8)));
storeValue(sp + 16, str);
setInt64(sp + 24, str.length);
},
// valueLoadString(v ref, b []byte)
"syscall/js.valueLoadString": (sp) => {
sp >>>= 0;
const str = loadValue(sp + 8);
loadSlice(sp + 16).set(str);
},
// func valueInstanceOf(v ref, t ref) bool
"syscall/js.valueInstanceOf": (sp) => {
sp >>>= 0;
this.mem.setUint8(sp + 24, (loadValue(sp + 8) instanceof loadValue(sp + 16)) ? 1 : 0);
},
// func copyBytesToGo(dst []byte, src ref) (int, bool)
"syscall/js.copyBytesToGo": (sp) => {
sp >>>= 0;
const dst = loadSlice(sp + 8);
const src = loadValue(sp + 32);
if (!(src instanceof Uint8Array || src instanceof Uint8ClampedArray)) {
this.mem.setUint8(sp + 48, 0);
return;
}
const toCopy = src.subarray(0, dst.length);
dst.set(toCopy);
setInt64(sp + 40, toCopy.length);
this.mem.setUint8(sp + 48, 1);
},
// func copyBytesToJS(dst ref, src []byte) (int, bool)
"syscall/js.copyBytesToJS": (sp) => {
sp >>>= 0;
const dst = loadValue(sp + 8);
const src = loadSlice(sp + 16);
if (!(dst instanceof Uint8Array || dst instanceof Uint8ClampedArray)) {
this.mem.setUint8(sp + 48, 0);
return;
}
const toCopy = src.subarray(0, dst.length);
dst.set(toCopy);
setInt64(sp + 40, toCopy.length);
this.mem.setUint8(sp + 48, 1);
},
"debug": (value) => {
console.log(value);
},
}
};
}
async run(instance) {
if (!(instance instanceof WebAssembly.Instance)) {
throw new Error("Go.run: WebAssembly.Instance expected");
}
this._inst = instance;
this.mem = new DataView(this._inst.exports.mem.buffer);
this._values = [ // JS values that Go currently has references to, indexed by reference id
NaN,
0,
null,
true,
false,
globalThis,
this,
];
this._goRefCounts = new Array(this._values.length).fill(Infinity); // number of references that Go has to a JS value, indexed by reference id
this._ids = new Map([ // mapping from JS values to reference ids
[0, 1],
[null, 2],
[true, 3],
[false, 4],
[globalThis, 5],
[this, 6],
]);
this._idPool = []; // unused ids that have been garbage collected
this.exited = false; // whether the Go program has exited
// Pass command line arguments and environment variables to WebAssembly by writing them to the linear memory.
let offset = 4096;
const strPtr = (str) => {
const ptr = offset;
const bytes = encoder.encode(str + "\0");
new Uint8Array(this.mem.buffer, offset, bytes.length).set(bytes);
offset += bytes.length;
if (offset % 8 !== 0) {
offset += 8 - (offset % 8);
}
return ptr;
};
const argc = this.argv.length;
const argvPtrs = [];
this.argv.forEach((arg) => {
argvPtrs.push(strPtr(arg));
});
argvPtrs.push(0);
const keys = Object.keys(this.env).sort();
keys.forEach((key) => {
argvPtrs.push(strPtr(`${key}=${this.env[key]}`));
});
argvPtrs.push(0);
const argv = offset;
argvPtrs.forEach((ptr) => {
this.mem.setUint32(offset, ptr, true);
this.mem.setUint32(offset + 4, 0, true);
offset += 8;
});
// The linker guarantees global data starts from at least wasmMinDataAddr.
// Keep in sync with cmd/link/internal/ld/data.go:wasmMinDataAddr.
const wasmMinDataAddr = 4096 + 8192;
if (offset >= wasmMinDataAddr) {
throw new Error("total length of command line and environment variables exceeds limit");
}
this._inst.exports.run(argc, argv);
if (this.exited) {
this._resolveExitPromise();
}
await this._exitPromise;
}
_resume() {
if (this.exited) {
throw new Error("Go program has already exited");
}
this._inst.exports.resume();
if (this.exited) {
this._resolveExitPromise();
}
}
_makeFuncWrapper(id) {
const go = this;
return function () {
const event = { id: id, this: this, args: arguments };
go._pendingEvent = event;
go._resume();
return event.result;
};
}
}
})();

8
shazam/fft.go
View file

@ -4,20 +4,20 @@ import (
"math" "math"
) )
// Fft performs the Fast Fourier Transform on the input signal. // FFT computes the Fast Fourier Transform (FFT) of the input data,
// converting the signal from the time domain to the frequency domain.
// For better understanding, refer to this video: https://www.youtube.com/watch?v=spUNpyF58BY
func FFT(input []float64) []complex128 { func FFT(input []float64) []complex128 {
// Convert input to complex128
complexArray := make([]complex128, len(input)) complexArray := make([]complex128, len(input))
for i, v := range input { for i, v := range input {
complexArray[i] = complex(v, 0) complexArray[i] = complex(v, 0)
} }
fftResult := make([]complex128, len(complexArray)) fftResult := make([]complex128, len(complexArray))
copy(fftResult, complexArray) // Copy input to result buffer copy(fftResult, complexArray)
return recursiveFFT(fftResult) return recursiveFFT(fftResult)
} }
// recursiveFFT performs the recursive FFT algorithm.
func recursiveFFT(complexArray []complex128) []complex128 { func recursiveFFT(complexArray []complex128) []complex128 {
N := len(complexArray) N := len(complexArray)
if N <= 1 { if N <= 1 {

36
shazam/filter.go
View file

@ -1,36 +0,0 @@
package shazam
import (
"math"
)
// LowPassFilter is a first-order low-pass filter using H(p) = 1 / (1 + pRC)
type LowPassFilter struct {
alpha float64 // Filter coefficient
yPrev float64 // Previous output value
}
// NewLowPassFilter creates a new low-pass filter
func NewLowPassFilter(cutoffFrequency, sampleRate float64) *LowPassFilter {
rc := 1.0 / (2 * math.Pi * cutoffFrequency)
dt := 1.0 / sampleRate
alpha := dt / (rc + dt)
return &LowPassFilter{
alpha: alpha,
yPrev: 0,
}
}
// Filter processes the input signal through the low-pass filter
func (lpf *LowPassFilter) Filter(input []float64) []float64 {
filtered := make([]float64, len(input))
for i, x := range input {
if i == 0 {
filtered[i] = x * lpf.alpha
} else {
filtered[i] = lpf.alpha*x + (1-lpf.alpha)*lpf.yPrev
}
lpf.yPrev = filtered[i]
}
return filtered
}

1
shazam/fingerprint.go
View file

@ -11,7 +11,6 @@ const (
) )
// Fingerprint generates fingerprints from a list of peaks and stores them in an array. // Fingerprint generates fingerprints from a list of peaks and stores them in an array.
// The fingerprints are encoded using a 32-bit integer format and stored in an array.
// Each fingerprint consists of an address and a couple. // Each fingerprint consists of an address and a couple.
// The address is a hash. The couple contains the anchor time and the song ID. // The address is a hash. The couple contains the anchor time and the song ID.
func Fingerprint(peaks []Peak, songID uint32) map[uint32]models.Couple { func Fingerprint(peaks []Peak, songID uint32) map[uint32]models.Couple {

3
shazam/image.go
View file

@ -11,11 +11,9 @@ import (
// ConvertSpectrogramToImage converts a spectrogram to a heat map image // ConvertSpectrogramToImage converts a spectrogram to a heat map image
func SpectrogramToImage(spectrogram [][]complex128, outputPath string) error { func SpectrogramToImage(spectrogram [][]complex128, outputPath string) error {
// Determine dimensions of the spectrogram
numWindows := len(spectrogram) numWindows := len(spectrogram)
numFreqBins := len(spectrogram[0]) numFreqBins := len(spectrogram[0])
// Create a new grayscale image
img := image.NewGray(image.Rect(0, 0, numFreqBins, numWindows)) img := image.NewGray(image.Rect(0, 0, numFreqBins, numWindows))
// Scale the values in the spectrogram to the range [0, 255] // Scale the values in the spectrogram to the range [0, 255]
@ -38,7 +36,6 @@ func SpectrogramToImage(spectrogram [][]complex128, outputPath string) error {
} }
} }
// Save the image to a PNG file
file, err := os.Create(outputPath) file, err := os.Create(outputPath)
if err != nil { if err != nil {
return err return err

90
shazam/shazam.go
View file

@ -1,3 +1,6 @@
//go:build !js && !wasm
// +build !js,!wasm
package shazam package shazam
import ( import (
@ -18,21 +21,35 @@ type Match struct {
Score float64 Score float64
} }
// FindMatches processes the audio samples and finds matches in the database // FindMatches analyzes the audio sample to find matching songs in the database.
func FindMatches(audioSamples []float64, audioDuration float64, sampleRate int) ([]Match, time.Duration, error) { func FindMatches(audioSample []float64, audioDuration float64, sampleRate int) ([]Match, time.Duration, error) {
startTime := time.Now() startTime := time.Now()
logger := utils.GetLogger()
spectrogram, err := Spectrogram(audioSamples, sampleRate) spectrogram, err := Spectrogram(audioSample, sampleRate)
if err != nil { if err != nil {
return nil, time.Since(startTime), fmt.Errorf("failed to get spectrogram of samples: %v", err) return nil, time.Since(startTime), fmt.Errorf("failed to get spectrogram of samples: %v", err)
} }
peaks := ExtractPeaks(spectrogram, audioDuration) peaks := ExtractPeaks(spectrogram, audioDuration)
fingerprints := Fingerprint(peaks, utils.GenerateUniqueID()) sampleFingerprint := Fingerprint(peaks, utils.GenerateUniqueID())
addresses := make([]uint32, 0, len(fingerprints)) sampleFingerprintMap := make(map[uint32]uint32)
for address := range fingerprints { for address, couple := range sampleFingerprint {
sampleFingerprintMap[address] = couple.AnchorTimeMs
}
matches, _, err := FindMatchesFGP(sampleFingerprintMap)
return matches, time.Since(startTime), nil
}
// FindMatchesFGP uses the sample fingerprint to find matching songs in the database.
func FindMatchesFGP(sampleFingerprint map[uint32]uint32) ([]Match, time.Duration, error) {
startTime := time.Now()
logger := utils.GetLogger()
addresses := make([]uint32, 0, len(sampleFingerprint))
for address := range sampleFingerprint {
addresses = append(addresses, address) addresses = append(addresses, address)
} }
@ -47,19 +64,34 @@ func FindMatches(audioSamples []float64, audioDuration float64, sampleRate int)
return nil, time.Since(startTime), err return nil, time.Since(startTime), err
} }
matches := map[uint32][][2]uint32{} // songID -> [(sampleTime, dbTime)] matches := map[uint32][][2]uint32{} // songID -> [(sampleTime, dbTime)]
timestamps := map[uint32][]uint32{} timestamps := map[uint32]uint32{} // songID -> earliest timestamp
targetZones := map[uint32]map[uint32]int{} // songID -> timestamp -> count
for address, couples := range m { for address, couples := range m {
for _, couple := range couples { for _, couple := range couples {
matches[couple.SongID] = append(matches[couple.SongID], [2]uint32{fingerprints[address].AnchorTimeMs, couple.AnchorTimeMs}) matches[couple.SongID] = append(
timestamps[couple.SongID] = append(timestamps[couple.SongID], couple.AnchorTimeMs) matches[couple.SongID],
[2]uint32{sampleFingerprint[address], couple.AnchorTimeMs},
)
if existingTime, ok := timestamps[couple.SongID]; !ok || couple.AnchorTimeMs < existingTime {
timestamps[couple.SongID] = couple.AnchorTimeMs
}
if _, ok := targetZones[couple.SongID]; !ok {
targetZones[couple.SongID] = make(map[uint32]int)
}
targetZones[couple.SongID][couple.AnchorTimeMs]++
} }
} }
// matches = filterMatches(10, matches, targetZones)
scores := analyzeRelativeTiming(matches) scores := analyzeRelativeTiming(matches)
var matchList []Match var matchList []Match
for songID, points := range scores { for songID, points := range scores {
song, songExists, err := db.GetSongByID(songID) song, songExists, err := db.GetSongByID(songID)
if !songExists { if !songExists {
@ -71,11 +103,7 @@ func FindMatches(audioSamples []float64, audioDuration float64, sampleRate int)
continue continue
} }
sort.Slice(timestamps[songID], func(i, j int) bool { match := Match{songID, song.Title, song.Artist, song.YouTubeID, timestamps[songID], points}
return timestamps[songID][i] < timestamps[songID][j]
})
match := Match{songID, song.Title, song.Artist, song.YouTubeID, timestamps[songID][0], points}
matchList = append(matchList, match) matchList = append(matchList, match)
} }
@ -86,7 +114,35 @@ func FindMatches(audioSamples []float64, audioDuration float64, sampleRate int)
return matchList, time.Since(startTime), nil return matchList, time.Since(startTime), nil
} }
// AnalyzeRelativeTiming checks for consistent relative timing and returns a score // filterMatches filters out matches that don't have enough
// target zones to meet the specified threshold
func filterMatches(
threshold int,
matches map[uint32][][2]uint32,
targetZones map[uint32]map[uint32]int) map[uint32][][2]uint32 {
// Filter out non target zones.
// When a target zone has less than `targetZoneSize` anchor times, it is not considered a target zone.
for songID, anchorTimes := range targetZones {
for anchorTime, count := range anchorTimes {
if count < targetZoneSize {
delete(targetZones[songID], anchorTime)
}
}
}
filteredMatches := map[uint32][][2]uint32{}
for songID, zones := range targetZones {
if len(zones) >= threshold {
filteredMatches[songID] = matches[songID]
}
}
return filteredMatches
}
// analyzeRelativeTiming calculates a score for each song based on the
// relative timing between the song and the sample's anchor times.
func analyzeRelativeTiming(matches map[uint32][][2]uint32) map[uint32]float64 { func analyzeRelativeTiming(matches map[uint32][][2]uint32) map[uint32]float64 {
scores := make(map[uint32]float64) scores := make(map[uint32]float64)
for songID, times := range matches { for songID, times := range matches {

136
shazam/shazamInit.go
View file

@ -1,136 +0,0 @@
package shazam
import (
"fmt"
"song-recognition/db"
"song-recognition/models"
"song-recognition/utils"
"sort"
)
type Match1 struct {
SongID uint32
SongTitle string
SongArtist string
YouTubeID string
Timestamp uint32
Coherency float64
}
func Search(audioSamples []float64, audioDuration float64, sampleRate int) ([]Match1, error) {
spectrogram, err := Spectrogram(audioSamples, sampleRate)
if err != nil {
return nil, fmt.Errorf("failed to get spectrogram of samples: %v", err)
}
peaks := ExtractPeaks(spectrogram, audioDuration)
fingerprints := Fingerprint(peaks, utils.GenerateUniqueID())
addresses := make([]uint32, 0, len(fingerprints))
for address, _ := range fingerprints {
addresses = append(addresses, address)
}
db, err := db.NewDBClient()
if err != nil {
return nil, err
}
defer db.Close()
couples, err := db.GetCouples(addresses)
if err != nil {
return nil, err
}
targetZones := targetZones(couples)
fmt.Println("TargetZones: ", targetZones)
matches := timeCoherency(fingerprints, targetZones)
var matchList []Match1
for songID, coherency := range matches {
song, songExists, err := db.GetSongByID(songID)
if err != nil || !songExists {
return nil, err
}
timestamp := targetZones[songID][0]
match := Match1{songID, song.Title, song.Artist, song.YouTubeID, timestamp, float64(coherency)}
matchList = append(matchList, match)
}
sort.Slice(matchList, func(i, j int) bool {
return matchList[i].Coherency > matchList[j].Coherency
})
return matchList, nil
}
func targetZones(m map[uint32][]models.Couple) map[uint32][]uint32 {
songs := make(map[uint32]map[uint32]int)
for _, couples := range m {
for _, couple := range couples {
if _, ok := songs[couple.SongID]; !ok {
songs[couple.SongID] = make(map[uint32]int)
}
songs[couple.SongID][couple.AnchorTimeMs]++
}
}
fmt.Println("couples: ", songs)
for songID, anchorTimes := range songs {
for msTime, count := range anchorTimes {
if count < 5 {
delete(songs[songID], msTime)
}
}
}
fmt.Println("anchorTimes: ", songs)
targetZones := make(map[uint32][]uint32)
for songID, anchorTimes := range songs {
for anchorTime, _ := range anchorTimes {
targetZones[songID] = append(targetZones[songID], anchorTime)
}
}
return targetZones
}
func timeCoherency(record map[uint32]models.Couple, songs map[uint32][]uint32) map[uint32]int {
// var threshold float64
matches := make(map[uint32]int)
for songID, songAnchorTimes := range songs {
deltas := make(map[float64]int)
for _, songAnchorTime := range songAnchorTimes {
for _, recordAnchor := range record {
recordAnchorTimeMs := float64(recordAnchor.AnchorTimeMs)
delta := recordAnchorTimeMs - float64(songAnchorTime)
deltas[delta]++
}
}
// Find the maximum number of time-coherent notes
var maxOccurrences int
for _, occurrences := range deltas {
if occurrences > maxOccurrences {
maxOccurrences = occurrences
}
}
matches[songID] = maxOccurrences
}
// Apply threshold for coherency
/**
for songID, coherency := range matches {
if float64(coherency) < threshold*float64(len(record)) {
delete(matches, songID) // Remove songs with insufficient coherency
}
}
*/
return matches
}

41
shazam/spectrogram.go
View file

@ -14,19 +14,17 @@ const (
hopSize = freqBinSize / 32 hopSize = freqBinSize / 32
) )
func Spectrogram(samples []float64, sampleRate int) ([][]complex128, error) { func Spectrogram(sample []float64, sampleRate int) ([][]complex128, error) {
lpf := NewLowPassFilter(maxFreq, float64(sampleRate)) filteredSample := LowPassFilter(maxFreq, float64(sampleRate), sample)
filteredSamples := lpf.Filter(samples)
downsampledSamples, err := Downsample(filteredSamples, sampleRate, sampleRate/dspRatio) downsampledSample, err := Downsample(filteredSample, sampleRate, sampleRate/dspRatio)
if err != nil { if err != nil {
return nil, fmt.Errorf("couldn't downsample audio samples: %v", err) return nil, fmt.Errorf("couldn't downsample audio sample: %v", err)
} }
numOfWindows := len(downsampledSamples) / (freqBinSize - hopSize) numOfWindows := len(downsampledSample) / (freqBinSize - hopSize)
spectrogram := make([][]complex128, numOfWindows) spectrogram := make([][]complex128, numOfWindows)
// Apply Hamming window function
window := make([]float64, freqBinSize) window := make([]float64, freqBinSize)
for i := range window { for i := range window {
window[i] = 0.54 - 0.46*math.Cos(2*math.Pi*float64(i)/(float64(freqBinSize)-1)) window[i] = 0.54 - 0.46*math.Cos(2*math.Pi*float64(i)/(float64(freqBinSize)-1))
@ -36,12 +34,12 @@ func Spectrogram(samples []float64, sampleRate int) ([][]complex128, error) {
for i := 0; i < numOfWindows; i++ { for i := 0; i < numOfWindows; i++ {
start := i * hopSize start := i * hopSize
end := start + freqBinSize end := start + freqBinSize
if end > len(downsampledSamples) { if end > len(downsampledSample) {
end = len(downsampledSamples) end = len(downsampledSample)
} }
bin := make([]float64, freqBinSize) bin := make([]float64, freqBinSize)
copy(bin, downsampledSamples[start:end]) copy(bin, downsampledSample[start:end])
// Apply Hamming window // Apply Hamming window
for j := range window { for j := range window {
@ -54,6 +52,29 @@ func Spectrogram(samples []float64, sampleRate int) ([][]complex128, error) {
return spectrogram, nil return spectrogram, nil
} }
// LowPassFilter is a first-order low-pass filter that attenuates high
// frequencies above the cutoffFrequency.
// It uses the transfer function H(s) = 1 / (1 + sRC), where RC is the time constant.
func LowPassFilter(cutoffFrequency, sampleRate float64, input []float64) []float64 {
rc := 1.0 / (2 * math.Pi * cutoffFrequency)
dt := 1.0 / sampleRate
alpha := dt / (rc + dt)
filteredSignal := make([]float64, len(input))
var prevOutput float64 = 0
for i, x := range input {
if i == 0 {
filteredSignal[i] = x * alpha
} else {
filteredSignal[i] = alpha*x + (1-alpha)*prevOutput
}
prevOutput = filteredSignal[i]
}
return filteredSignal
}
// Downsample downsamples the input audio from originalSampleRate to targetSampleRate // Downsample downsamples the input audio from originalSampleRate to targetSampleRate
func Downsample(input []float64, originalSampleRate, targetSampleRate int) ([]float64, error) { func Downsample(input []float64, originalSampleRate, targetSampleRate int) ([]float64, error) {
if targetSampleRate <= 0 || originalSampleRate <= 0 { if targetSampleRate <= 0 || originalSampleRate <= 0 {

64
wasm/wasm_main.go Normal file
View file

@ -0,0 +1,64 @@
//go:build js && wasm
// +build js,wasm
package main
import (
"song-recognition/shazam"
"song-recognition/utils"
"syscall/js"
)
func generateFingerprint(this js.Value, args []js.Value) interface{} {
if len(args) < 2 {
return js.ValueOf(map[string]interface{}{
"error": 1,
"data": "Expected audio array and sample rate",
})
}
if args[0].Type() != js.TypeObject || args[1].Type() != js.TypeNumber {
return js.ValueOf(map[string]interface{}{
"error": 2,
"data": "Invalid argument types; Expected audio array and samplerate (type: int)",
})
}
inputArray := args[0]
sampleRate := args[1].Int()
audioData := make([]float64, inputArray.Length())
for i := 0; i < inputArray.Length(); i++ {
audioData[i] = inputArray.Index(i).Float()
}
spectrogram, err := shazam.Spectrogram(audioData, sampleRate)
if err != nil {
return js.ValueOf(map[string]interface{}{
"error": 3,
"data": "Error generating spectrogram: " + err.Error(),
})
}
peaks := shazam.ExtractPeaks(spectrogram, float64(len(audioData)/sampleRate))
fingerprint := shazam.Fingerprint(peaks, utils.GenerateUniqueID())
fingerprintArray := []interface{}{}
for address, couple := range fingerprint {
entry := map[string]interface{}{
"address": address,
"anchorTime": couple.AnchorTimeMs,
}
fingerprintArray = append(fingerprintArray, entry)
}
return js.ValueOf(map[string]interface{}{
"error": 0,
"data": fingerprintArray,
})
}
func main() {
js.Global().Set("generateFingerprint", js.FuncOf(generateFingerprint))
select {}
}