dtmf-tone-pure

Zero-dependency DTMF tone generator — pure stdlib Python, RFC 4733-compliant WAV output.

Generate all 16 ITU Q.23/Q.24 DTMF tones as WAV bytes using only math and struct. No numpy, no C extensions, no dependencies. Designed for AWS Lambda, Cloudflare Workers, and embedded Python.

Quick Start

pip install dtmf-tone-pure

from dtmf_tone_pure import DTMFTone

tone = DTMFTone().tone('5')       # Single tone, 44+ bytes of WAV
tones = DTMFTone().tones('911')   # Multi-digit with 10ms inter-digit pause
coeff = DTMFTone().goertzel_coefficient(697, block_size=120)  # Goertzel coefficient

Install from source:

pip install git+https://github.com/prasad-a-abhishek/dtmf-tone-pure.git

Run tests:

pip install -e .
pytest tests/ -v

Why dtmf-tone-pure?

Every existing DTMF library for Python depends on numpy or C extensions. In serverless environments (AWS Lambda cold start ~50ms for numpy import, ~50MB memory), that's prohibitive.

dtmf-tone-pure fills the gap: the only zero-dependency Python package on PyPI that generates DTMF tones as RFC 4733-compliant WAV bytes. (pydtmf on PyPI does detection only.)

Key Features

• 16 DTMF tones: 0–9, A–D, *, # per ITU Q.23/Q.24
• Zero dependencies: pure stdlib — math, struct, typing only
• WAV output: 8 kHz, 16-bit mono PCM with correct RIFF header
• Goertzel coefficient: goertzel_coefficient(freq, block_size) matching analytical values
• ITU-spec durations: 60ms tone, 10ms inter-digit pause
• Subclassable: override FREQ_TABLE for custom frequency pairs

API Reference

from dtmf_tone_pure import DTMFTone

dtmf = DTMFTone()

# Class constants
dtmf.SAMPLE_RATE      # 8000 Hz
dtmf.TONE_DURATION    # 0.060 s (60 ms)
dtmf.PAUSE_DURATION   # 0.010 s (10 ms)
dtmf.FREQ_TABLE       # {'1': (697.0, 1209.0), ...} — 16 entries

# Methods
dtmf.tone(key: str) -> bytes
    # Single DTMF tone as WAV bytes.
    # key: '0'-'9', 'A'-'D', '*', '#'
    # Returns: 44-byte RIFF WAV header + PCM samples

dtmf.tones(digits: str) -> bytes
    # Concatenated tones with 10ms inter-digit silence.
    # Returns: single valid WAV file

dtmf.goertzel_coefficient(freq: float, *, block_size: int = 120) -> float
    # Goertzel coefficient = 2 * cos(2π * k / N)
    # k = round(N * freq / sample_rate)
    # Returns: float rounded to 6 decimal places

DTMF Frequency Table (ITU Q.23/Q.24):

Limitations

• Single sample rate (8000 Hz). Custom rates require subclassing and recomputing.
• Amplitude is fixed at 16383.5 scale factor (max sum-of-two-sines → max int16).
• tones() returns a single concatenated WAV; no gapless-playback marker between segments.
• Goertzel coefficient precision is rounded to 6 decimal places.

Non-Goals

• CLI interface (see goertzel-tone-pure CLI wrapper if needed)
• Tone detection (use pydtmf for that)
• numpy/scipy acceleration
• Multiple sample rates or bit depths beyond 8kHz/16-bit mono

License

MIT — see LICENSE.