TuneTracer's Algorithm

1. WAV Loading & Preprocessing

• Raw PCM extraction: We parse the RIFF/WAVE header ourselves (no external libraries) to load 16 bit PCM samples.
• Spectrogram (STFT): Audio is framed into 2048-sample windows (≈46 ms) with 50% overlap, a Hann window is applied, and a radix-2 FFT is run to produce a high-resolution magnitude spectrogram (S[t,f]).

2. Peak Detection (“Constellation”)

• Local maxima: For each time frame and frequency bin, we detect peaks that exceed a dynamic threshold (mean + 1.5 σ) and are the highest in a 7×7 neighborhood.
• Node list: Each peak (p_i = (t_i, f_i, A_i)) becomes a node in our fingerprint graph.

3. Graph-Based Descriptors

• Neighbor graph: For each peak (p_i), we connect it to up to K = 5 future peaks (p_j) within ±Δf (e.g. ±50 bins) and Δt (up to 1 s).
• Amplitude weighting: Each edge is weighted by (A_i \times A_j), so louder events carry more significance.
• Local descriptor: We record ((\Delta t, \Delta f, w)) for each of the K strongest edges, forming a 3×K vector around each anchor.

4. Quantization & Hashing

• Bins: (\Delta t) quantized into 16 time-bins, (\Delta f) into 32 freq-bins, and (w) into 8 amplitude-bins.
• FNV-1a hashing: The 3K bytes are concatenated and hashed into a 64-bit fingerprint code, compactly encoding subgraph structure.

5. Persistent Indexing

• All (code, songId, timestamp) tuples are bulk‐inserted into SQLite with an index on code.
• This on-disk database scales to millions of fingerprints and requires no external key-value service.

6. Query & Matching

1. Fingerprint snippet: The same STFT → peaks → graph → hashes pipeline runs on the query.
2. Lookup: Each code is looked up in SQLite, returning (songId, dbTimestamp) entries.
3. Voting: We tally votes for each (songId, Δt = queryTime – dbTime) pair.
4. Best match: The song and time-offset with the highest vote count wins.

Why Tune Tracer Beats Classic Shazam

• Richer descriptors: Captures a small subgraph of K edges per anchor vs. a single peak pair.
• Amplitude-aware: Weighs edges by peak magnitudes for robustness to noise.
• Finer quantization: More granular time/frequency bins for higher discrimination.