diff --git a/shazam/shazam.go b/shazam/shazam.go
index de9d5fd..52a90cc 100644
--- a/shazam/shazam.go
+++ b/shazam/shazam.go
@@ -3,7 +3,6 @@ package shazam
 import (
 	"fmt"
 	"math"
-	"song-recognition/models"
 	"song-recognition/utils"
 	"sort"
 )
@@ -17,6 +16,7 @@ type Match struct {
 	Score      float64
 }
 
+// FindMatches processes the audio samples and finds matches in the database
 func FindMatches(audioSamples []float64, audioDuration float64, sampleRate int) ([]Match, error) {
 	logger := utils.GetLogger()
 
@@ -27,10 +27,9 @@ func FindMatches(audioSamples []float64, audioDuration float64, sampleRate int)
 
 	peaks := ExtractPeaks(spectrogram, audioDuration)
 	fingerprints := Fingerprint(peaks, utils.GenerateUniqueID())
-	fmt.Println("peaks len: ", len(peaks))
 
 	addresses := make([]uint32, 0, len(fingerprints))
-	for address, _ := range fingerprints {
+	for address := range fingerprints {
 		addresses = append(addresses, address)
 	}
 
@@ -45,34 +44,17 @@ func FindMatches(audioSamples []float64, audioDuration float64, sampleRate int)
 		return nil, err
 	}
 
-	matches := map[uint32]map[uint32]models.Couple{}
+	matches := map[uint32][][2]uint32{} // songID -> [(sampleTime, dbTime)]
 	timestamps := map[uint32]uint32{}
 
 	for address, couples := range m {
 		for _, couple := range couples {
-
-			if _, ok := matches[couple.SongID]; !ok {
-				matches[couple.SongID] = map[uint32]models.Couple{}
-				timestamps[couple.SongID] = couple.AnchorTimeMs
-			}
-
-			matches[couple.SongID][address] = couple
+			matches[couple.SongID] = append(matches[couple.SongID], [2]uint32{fingerprints[address].AnchorTimeMs, couple.AnchorTimeMs})
+			timestamps[couple.SongID] = couple.AnchorTimeMs
 		}
 	}
 
-	scores := map[uint32]float64{}
-	for songID, couples := range matches {
-		song, songExists, err := db.GetSongByID(songID)
-		if err != nil || !songExists {
-			// log error
-			fmt.Println("Continuing")
-			continue
-		}
-		fmt.Printf("Song: %v, Scores:\n", song.Title)
-
-		scores[songID] = matchScore(fingerprints, couples)
-		fmt.Println("------------------------------------")
-	}
+	scores := analyzeRelativeTiming(matches)
 
 	var matchList []Match
 	for songID, points := range scores {
@@ -87,7 +69,6 @@ func FindMatches(audioSamples []float64, audioDuration float64, sampleRate int)
 		}
 
 		fmt.Printf("Song: %v, Score: %v\n", song.Title, points)
-		fmt.Println("====================================")
 		match := Match{songID, song.Title, song.Artist, song.YouTubeID, timestamps[songID], points}
 		matchList = append(matchList, match)
 	}
@@ -100,60 +81,21 @@ func FindMatches(audioSamples []float64, audioDuration float64, sampleRate int)
 	return matchList, nil
 }
 
-// MatchScore computes a match score between the two transformed audio samples (into a list of Key + TableValue)
-func matchScore(sample, match map[uint32]models.Couple) float64 {
-	// Will hold a list of points (time in the sample sound file, time in the matched database sound file)
-	points := [2][]float64{}
-	matches := 0.0
-	for k, sampleValue := range sample {
-		if matchValue, ok := match[k]; ok {
-			points[0] = append(points[0], float64(sampleValue.AnchorTimeMs))
-			points[1] = append(points[1], float64(matchValue.AnchorTimeMs))
-			matches++
+// AnalyzeRelativeTiming checks for consistent relative timing and returns a score
+func analyzeRelativeTiming(matches map[uint32][][2]uint32) map[uint32]float64 {
+	scores := make(map[uint32]float64)
+	for songID, times := range matches {
+		count := 0
+		for i := 0; i < len(times); i++ {
+			for j := i + 1; j < len(times); j++ {
+				sampleDiff := math.Abs(float64(times[i][0] - times[j][0]))
+				dbDiff := math.Abs(float64(times[i][1] - times[j][1]))
+				if math.Abs(sampleDiff-dbDiff) < 100 { // Allow some tolerance
+					count++
+				}
+			}
 		}
+		scores[songID] = float64(count)
 	}
-	corr := correlation(points[0], points[1])
-	fmt.Printf("Score (%v * %v * %v): %v\n", corr, corr, matches, corr*corr*matches)
-	return corr * corr * matches
-}
-
-// Correlation computes the correlation between 2 series of points
-// the length used is x's
-func correlation(x []float64, y []float64) float64 {
-	n := len(x)
-	meanX, meanY := Avg(x[:n]), Avg(y[:n])
-
-	sXY := 0.0
-	sX := 0.0
-	sY := 0.0
-
-	for i, xp := range x {
-		dx := xp - meanX
-		dy := y[i] - meanY
-
-		sX += dx * dx
-		sY += dy * dy
-
-		sXY += dx * dy
-	}
-
-	if sX == 0 || sY == 0 {
-		return 0
-	}
-
-	return sXY / (math.Sqrt(sX) * math.Sqrt(sY))
-}
-
-// Avg computes the average of the given array
-func Avg(arr []float64) float64 {
-	if len(arr) == 0 {
-		return 0
-	}
-
-	sum := 0.0
-	for _, v := range arr {
-		sum += v
-	}
-
-	return sum / float64(len(arr))
+	return scores
 }