Tape hiss, traffic and AC noise, even old phone lines. Sound Cleaning removes ambient noise and echo while preserving the voice as it was spoken. One button, one file.
Rashid bin Tannaf bin Fayez bin Saeed was born in 1910 in Al Dhaid, Sharjah. He lost his father early, was raised by his maternal grandfather, and memorised the Quran before he could write. At seventeen he went to sea as a pearl diver and stayed thirty years; later he quarried coral stone for houses. He turned to poetry at twenty-five.
His Nabati verse was half desert, half coast, with a port's worth of English words slipped in, and it was funny. Wisdom carried by a joke. But what made him famous was the voice: the delivery, the pause, the punchline dropped between two lines. Radio and television carried it through the 1970s and 80s, and his majlis in Al Ramla, Sharjah, filled every night.
He published Sada al-Wona in 1984 and Ahwak in 1997, followed by Diwan Tannaf in 2018. In 2009, Emirates Post issued a stamp bearing his image. He died in 2000, and most of his legacy survives on tape: his words as he spoke them, in his voice, with all the surrounding noise. This page preserves his voice and removes that noise.
* Facts checked against Al Ittihad, Al Khaleej, Al Bayan, Emirates Post and Arabic Wikipedia. Sources disagree on tribe (Al Kaabi / Al Nuaimi) and death year (1999–2001).
Most audio cleaning tools give you one slider and a promise of improvement. But sound engineers, podcasters, archivists and voice artists want to know what actually changed. Advanced cleaning reveals the eight settings behind each preset, from low-frequency reduction to loudness and dynamic range. Each change appears in red, and you hear its effect on the raw recording as you move the slider.
Developed with professionals who spend their day wearing headphones. Set the target loudness for your podcast or platform. Raise the low-frequency filter to counter wind noise, or lower it to preserve the depth of a voice. Control the dynamic range for a quiet reading or a majlis with overlapping voices. Your ears are the judge.
Press ▶ Hear it, then move any slider. This line tells you what the change does to the tape.
The dots represent the sound, frequency band by frequency band: low frequencies on the left, high frequencies on the right. The Raw recording row shows the tape as recorded, while the Cleaned audio row shows the result after applying your settings. Use Raw recording / Your settings to switch between the two.
Pick the room the recording came from. The model removes what does not belong and keeps the voice whole: breath, pauses, the dialect's own rhythm. Up to 1 hour per file.
Podcasts, voice-overs, interviews. Removes hiss and room tone, leaves everything human.
Air conditioning, wind on the mic, traffic behind a site interview.
Narrow, crackling sources: call recordings, archive cassettes, radio pulls.
Tuned so a deep speaker does not come back thin after cleaning.
Sound Cleaning sits in front of the other products. Same login, same file, one step earlier.
A cleaned file gives Speech to Text a clearer signal, turning noisy calls into accurate transcripts ready to use.
Voice Cloning learns from what it hears. Give it the voice, not the room.
Voice notes, site visits, majlis recordings, archive tape. Publish them without apologising for the sound.
Send the audio file, choose a preset and receive the cleaned version. You can pass it directly to transcription within the same API request.
Coming Soon# python, a preview of the Sound Cleaning API from echonova import EchoNova client = EchoNova() clean = client.audio.clean( audio="tannaf_1980s.mp3", preset="radio", # studio, wind, radio or deep ) clean.save("tannaf_clean.wav") text = client.stt.transcribe(clean)
Security & sovereignty
Speech to TextConvert audio to text in real time or in batches, across eight dialects.
Text to SpeechNatural Arabic voices in the caller’s own dialect.
Noise and echo are removed before the model listens to the audio.
From calls, apps or recordings.
Choose the preset that matches the recording conditions and compare the output with the original audio.
Cleaning produces processed audio. Speech to Text is a separate tool that produces a transcript.
Sovereign Arabic voice AI
