grundig-dss 1.0.0

Bit-exact native decoder for Grundig DSS-SP (PH9607) dictation audio — .dss to 16 kHz WAV, no Windows and no vendor software.

Grundig DSS-SP native decoder

Pure-Python, bit-exact decoder for Grundig .dss files (magic \x06dss, e.g. the Digta 415) — the Grundig SP CELP variant that neither FFmpeg's dss_sp, the Olympus-derived open codecs, nor NCH Switch decode correctly.

Reverse-engineered from Grundig's dss2wav.dll; output is byte-for-byte identical to the reference on every test sample. The full story and the recovered bitstream + synthesis spec are in ../docs/12-cracking-the-grundig-sp-codec.md.

Install

pip install grundig-dss        # from PyPI (pure-Python, zero dependencies)

Usage

grundig-dss input.dss [output.wav]    # console command (out defaults to the input stem)
# or, without installing:
python3 -m grundig input.dss output.wav
./decode.sh input.dss output.wav

from grundig import decode_dss, write_wav
pcm = decode_dss("recording.dss")     # list[int], 16 kHz mono 16-bit
write_wav("recording.wav", pcm)

Files

File	What
grundig_dss.py	the decoder (no Wine, no DLL, stdlib only)
gtables.json	quantization tables (reflection codebooks, gains, pulse/binomial tables, resample FIR) extracted from dss2wav.dll's .data
decode.sh	thin CLI wrapper

Olympus DSS/DS2 files (\x02dss/\x03dss/\x03ds2) are handled by the main chain, not this module.