فارسی | English
Transfer text over sound - A Progressive Web App that encodes text into audio signals ("Neda") that can be transmitted between devices over a phone call or through speakers to microphones. Works completely without internet after the first visit.
"Neda" is a Persian feminine given name meaning "voice," "call," or "calling"
Nedagram was built to help share text data when sending files isn't possible. By converting text to audio, you can:
- Share config files, complex passwords, or long URLs over phone calls when internet is unavailable
- Transfer data between air-gapped devices (private keys, signatures)
- Off-grid communication - works anywhere with no network required
- Fully offline after first load - no server, no tracking, no data leaves your device
┌─────────────┐ Audio ┌─────────────┐
│ SENDER │ Waves │ RECEIVER │
│ │ ~~~~~~~~> │ │
│ Text Input │ Speaker │ Microphone │
│ ↓ │ to │ ↓ │
│ Encrypt* │ Mic │ Decode │
│ ↓ │ (phone) │ ↓ │
│ Compress │ │ FEC Decode │
│ ↓ │ │ ↓ │
│ FEC Encode │ │ Decompress │
│ ↓ │ │ ↓ │
│ MFSK Audio │ │ Decrypt* │
└─────────────┘ └─────────────┘
* if encrypted
- Sender enters text (config file, password, URL, etc.)
- Optionally encrypts data with a password
- App compresses and encodes data with error correction
- Data is modulated into audio tones (MFSK)
- Receiver captures audio via microphone
- App decodes, corrects errors, and decrypts if needed
- Both parties verify SHA-256 checksum matches
- Offline-First PWA - Works without internet after first load
- End-to-End Encryption - Optional ChaCha20-Poly1305 encryption with password protection
- Two Audio Modes
- Phone Mode - Works over standard phone calls (300-3400 Hz)
- Wideband Mode - Faster transmission for direct device-to-device or HD Voice
- Error Correction - Concatenated FEC (Reed-Solomon + Convolutional with Viterbi decoding)
- Auto-Detection - Receiver automatically detects transmission mode
- Compression - DEFLATE compression reduces transmission time
- Integrity Verification - SHA-256 checksum for sender/receiver verification
- QR Code Fallback - For small payloads (< 2KB)
- Multi-Language - English and Farsi (RTL) support
- Downloadable Offline Package - Share the app itself without internet
- Share config files, complex passwords, or long URLs over phone calls when internet is unavailable
- Transfer data between air-gapped devices (private keys, signatures)
- Off-grid communication - works anywhere with no network required
- Fully offline after first load
- No server, no tracking, no data leaves your device
- Open Nedagram on the sending device
- Paste your text or upload a file
- Select Phone (for calls) or Wideband (for direct transfer)
- Optionally enable Encrypt and enter a password
- Click Generate Audio
- Play the audio near the receiving device
- Share the SHA-256 checksum (and password if encrypted) with the receiver
- Open Nedagram on the receiving device
- Click Start Listening
- Allow microphone access
- Wait for the transmission to complete
- If encrypted, enter the password to decrypt
- Verify the checksum matches the sender's
- Copy or save the decoded text
| Parameter | Value |
|---|---|
| Modulation | 4-MFSK (2 bits/symbol) |
| Tone Frequencies | 800, 1300, 1800, 2300 Hz |
| Tone Spacing | 500 Hz (wide for codec tolerance) |
| Symbol Duration | 50ms + 12ms guard |
| Effective Bitrate | ~20-25 bps |
| Synchronization | Chirp matched filter detection |
| Burst Protection | Block interleaving enabled |
| Parameter | Value |
|---|---|
| Modulation | 16-MFSK (4 bits/symbol) |
| Frequency Range | 1800 - 5700 Hz |
| Symbol Duration | 40ms + 5ms guard |
| Effective Bitrate | ~50-60 bps |
Same scheme used by NASA's Voyager spacecraft for reliable deep space communication.
| Parameter | Value |
|---|---|
| Outer Code | Reed-Solomon with 16 parity bytes |
| Inner Code | Convolutional (k=7, rate 2/3 punctured) |
| Viterbi Decoding | Soft-decision, 35-symbol traceback |
| Scrambling | LFSR (x^15 + x^14 + 1) for bit distribution |
| Error Correction | ~8 byte errors + additional bit errors per frame |
| Component | Phone Mode | Wideband Mode |
|---|---|---|
| Warmup Tone | 200ms steady tone | 400ms steady tone |
| Chirp Sweep | 800ms (600-2600 Hz) | 1200ms (1000-4000 Hz) |
| Detection | Matched filter cross-correlation | |
| Calibration | 4 tones × 2 repeats | 4 tones × 3 repeats |
| Sync Pattern | 8-symbol alternating pattern |
| Parameter | Value |
|---|---|
| Cipher | ChaCha20-Poly1305 (AEAD) |
| Key Derivation | PBKDF2-SHA256 (100,000 iterations) |
| Overhead | 44 bytes (16 salt + 12 nonce + 16 auth tag) |
- Maximum payload: 100 KB
- Recommended payload: < 50 KB
- QR code available for payloads < 2 KB
| Payload Size | Phone Mode | Wideband Mode |
|---|---|---|
| 100 bytes | ~45 sec | ~20 sec |
| 1 KB | ~6 min | ~2.5 min |
| 10 KB | ~55 min | ~25 min |
Times are approximate and include preamble, headers, and FEC overhead. Phone mode is optimized for reliability over speed.
- Quiet environment - Background noise reduces accuracy
- Volume at 70-80% - Too loud causes distortion, too quiet loses signal
- Distance 0.5-1m - Optimal range for speaker-to-microphone
- Keep devices steady - Movement during transmission can cause errors
- Verify checksum - Always compare SHA-256 to confirm integrity
You can download Nedagram for completely offline use - no internet connection needed after the initial download.
Visit the Help page at nedagram.com and click "Download Offline Package" to get a ZIP file containing everything you need.
Direct download: nedagram.com/nedagram-offline.zip
# Clone and build
git clone https://github.com/shayanb/nedagram.git
cd nedagram
npm install
npm run build
# The zip file will be at dist/nedagram-offline.zip- Extract the ZIP file to a folder
- Start a local web server:
# Using Python (recommended) python3 -m http.server 8000 --bind 127.0.0.1 # Using Node.js npx serve . # Using PHP php -S 127.0.0.1:8000
- Open http://127.0.0.1:8000 in your browser
Important: Use
http://127.0.0.1:8000(not[::]orlocalhost) for microphone access to work. For other devices on your network, use your computer's IP address (e.g.,http://192.168.1.x:8000).
The offline package includes the CLI in the cli/ folder. Run it directly with Node.js:
node cli/index.cjs encode "Hello" -o message.wav
node cli/index.cjs decode message.wavNedagram includes a CLI for encoding and decoding WAV files without a browser.
# Install globally from npm
npm install -g nedagram
# Or run directly with npx
npx nedagram --help
# Or use from the offline package (no install needed)
node cli/index.cjs --help# Encode text to WAV file
nedagram encode "Hello World" -o message.wav
# Encode from file
nedagram encode -f config.txt -o config.wav
# Encode with encryption
nedagram encode "Secret data" -o encrypted.wav -e -p "password"
# Encode with phone mode (for phone calls)
nedagram encode "Text" -o phone.wav -m phone
# Decode WAV file
nedagram decode message.wav
# Decode encrypted file
nedagram decode encrypted.wav -p "password"
# Decode to file
nedagram decode message.wav -o output.txt
# Pipe input/output
echo "Hello" | nedagram encode -o hello.wav
nedagram decode message.wav > output.txtEncode:
| Option | Description |
|---|---|
-f, --file <path> |
Read input from file |
-o, --output <path> |
Output WAV file path |
-m, --mode <mode> |
Audio mode: phone or wideband (default) |
-e, --encrypt |
Encrypt the message |
-p, --password <pwd> |
Password for encryption |
-q, --quiet |
Suppress progress output |
--json |
Output result as JSON (includes metadata, sha256) |
Decode:
| Option | Description |
|---|---|
-o, --output <path> |
Write decoded text to file |
-p, --password <pwd> |
Password for decryption |
-q, --quiet |
Suppress progress output |
--json |
Output result as JSON (includes message, metadata, sha256) |
- Node.js 18+
- npm
# Clone repository
git clone https://github.com/shayanb/nedagram.git
cd nedagram
# Install dependencies
npm install
# Start development server (with HTTPS for mic access)
npm run dev
# Build for production
npm run build
# Build CLI
npm run build:cli
# Run CLI (use -- to pass arguments through npm)
npm run cli -- encode "Hello" -o test.wav
# Or run directly:
./dist-cli/nedagram-cli/index.cjs encode "Hello" -o test.wav
# Run tests
npm test
# Run CLI tests
npm run test:cli- Framework: Preact + TypeScript
- Build Tool: Vite
- State: Preact Signals
- Compression: pako (DEFLATE)
- Encryption: @noble/ciphers (ChaCha20-Poly1305)
- Error Correction: Concatenated FEC (Reed-Solomon + Convolutional/Viterbi)
- PWA: Service Worker with offline caching
- Encryption - Always protect sensitive data with a password
- Offline capable - Works without any network after first load, also has a downloadable offline package (& CLI)
- No telemetry - No data is sent anywhere
- No backend - Entirely client-side
- Open source - Audit the code yourself, please do!
| Browser | Support |
|---|---|
| Chrome (Desktop/Android) | Full |
| Firefox | Full |
| Safari (macOS/iOS) | Full |
| Edge | Full |
Requires microphone permission for receiving.
If you encounter issues with decoding or any other problems, please report them on GitHub:
- Click the debug log button (bottom-right corner of the app) to enable logging
- Reproduce the issue
- Click "Copy" in the debug panel to copy the log
- If possible, click "Save Audio" to save the recorded audio file
- Create a GitHub issue with:
- The debug log
- The audio file (if available)
- Description of what you were trying to do
Contributions are welcome! Please feel free to submit issues and pull requests.
MIT License - See LICENSE for details.
This software is provided "as is" for educational and research purposes. See DISCLAIMER.md for full terms.
Note: This code has not been audited. The audio encoding/decoding implementation is based on established research papers and popular codebases in the field, developed with AI assistance and guided by the author(s). Use at your own risk for sensitive applications.
Built with the goal of helping people communicate freely when traditional methods fail.
Nedagram - When internet fails, sound finds a way. 📞