AI Voice-to-Text Bot Zeeno Voice
Built 0-to-100 by Bizeeno — an intelligent AI speech-to-text Telegram bot powered by Whisper AI. Converts voice messages, podcasts, and audio attachments into accurate Persian & English text in under 3 seconds.
Key Performance Metrics
Processing speed, accuracy rate, and user feedback metrics:
Interactive Voice Transcription Demo
Select a sample audio file below to see instant AI text transcription:
Sample Audio Notes:Click to test
Engine automatically suppresses street noise, mic hiss, and ambient static before serving clean audio to Whisper AI.
Hello! Following up on yesterday’s meeting, phase 1 of the website should be ready by the end of the week, and we will perform final tests on the AI algorithms. Please review the progress report.
Key Features & Engineering Values
1. 100% Free & Built by Bizeeno
All software, server processing, and Telegram bot architecture were developed in-house by Bizeeno and offered completely free to all users.
- No subscriptions or sign-up fees
- Designed 0-to-100 by Bizeeno engineering team
2. Powered by Whisper AI Speech Recognition
Utilizes Whisper neural network models for high Persian & English word accuracy, handling technical terms and proper punctuation.
- 98.5% word accuracy rate for Persian voice notes
- Dual Persian & English multi-lingual support
3. FFmpeg Pre-processing & Noise Reduction
Audio files are filtered via FFmpeg before hitting the AI model, removing ambient static, traffic sound, and weak microphone noise.
- Full support for OGG, MP3, WAV and Telegram audio formats
- Sampling rate optimization for sub-3s response speed
4. Instant Speed & Easy Access
Simply forward any audio note to the Telegram bot to get complete text back in <3 seconds without installing app dependencies.
- Instant voice transcription with zero length limits
- Cross-device access on mobile, tablet & desktop Telegram
AI Pipeline & Architecture (Zeeno STT Engine)
Simulation of Python async pipeline, FFmpeg audio filtering & Whisper STT
# Zeeno Voice Bot - Async AI Speech-to-Text Pipeline (Python + Whisper AI)
import asyncio
import ffmpeg
import whisper
from telegram import Update
from telegram.ext import ApplicationBuilder, MessageHandler, filters, ContextTypes
# Load OpenAI Whisper STT Model into GPU memory
stt_model = whisper.load_model("small")
async def process_voice_message(update: Update, context: ContextTypes.DEFAULT_TYPE):
user_id = update.effective_user.id
voice_file = await update.message.voice.get_file()
raw_path = f"/tmp/zeeno_raw_{user_id}.ogg"
clean_path = f"/tmp/zeeno_clean_{user_id}.wav"
await voice_file.download_to_drive(raw_path)
# Noise reduction & audio normalization with FFmpeg
ffmpeg.input(raw_path).output(clean_path, ar="16000", ac=1).run(overwrite_output=True)
# Transcribe Persian/English voice with Whisper AI
result = stt_model.transcribe(clean_path, language="fa", fp16=False)
extracted_text = result.get("text", "").strip()
await update.message.reply_text(f"📝 متن استخراجشده:\n\n{extracted_text}")Want your custom bot or AI-powered system built?
Bizeeno’s AI and software team is ready to build your custom Telegram bots, AI speech systems, and voice automation tools.