OpenTone

A data-over-sound toolkit in pure Python. OpenTone carries data through an audio channel using a family of schemes: telephone tones, modems, Morse, image transmission, and watermarking. Each scheme is its own submodule that shares one set of WAV, synthesis, and Goertzel primitives. The core schemes have no dependencies.

Install

pip install opentone            # core schemes (no dependencies)
pip install opentone[all]       # + image/DSP schemes (numpy, Pillow)

Schemes

Quickstart

DTMF (also re-exported at the top level for backwards compatibility):

from opentone import encode_text, decode

encode_text("hello world", "message.wav")
print(decode("message.wav"))          # "hello world"

Morse:

from opentone import morse
morse.encode_text("SOS", "sos.wav")
print(morse.decode("sos.wav"))        # "SOS"

FSK modem, with optional Reed-Solomon error correction:

from opentone import fsk
fsk.encode(b"arbitrary bytes", "data.wav", nsym=8)
print(fsk.decode("data.wav", nsym=8))  # b"arbitrary bytes"

Caller-ID:

from opentone import callerid
callerid.encode("call.wav", number="5551234567", name="JARBAS AI",
                timestamp="06041530", mdmf=True)
print(callerid.decode("call.wav"))

Image into a spectrogram, and slow-scan-TV image transmission:

from opentone import spectrogram, sstv
spectrogram.encode_image("logo.png", "logo.wav")   # view the WAV's spectrogram
sstv.encode_image("photo.png", "photo.wav", width=160, height=120)
img = sstv.decode_image("photo.wav", width=160, height=120)

Hide data in existing audio:

from opentone import watermark
watermark.embed("song.wav", "marked.wav", b"owner-id")
print(watermark.extract("marked.wav", 8))   # b"owner-id"

Input audio

The tone schemes write 8 kHz mono 16-bit WAV. To decode audio from another source, convert it first:

ffmpeg -i some_file.mp3 -acodec pcm_s16le -ac 1 -ar 8000 out.wav

Documentation

• Quickstart
• DTMF: encoding · decoding · reference
• Morse · FSK modem · Caller-ID · MF signalling
• Error correction
• SSTV · Spectrogram art · Watermarking
• API reference

Runnable scripts are in examples/.

License

MIT