PythonWhisper APIGPT-4oElevenLabs TTS
The problem
Real-time voice translation either costs a fortune (Zoom AI) or sounds robotic (Google Translate). Nothing Telegram-native for async voice messages with a client abroad.
Before / After
What changed
Before
- Record voice, save, upload to translator, copy text
- Paste into ChatGPT, generate response
- Paste into a TTS service, download audio, forward
- 90 seconds of context-switching per message
After
- Send a voice message to @clappix_voice_bot
- Whisper transcribes → GPT translates → TTS synthesizes
- Get an English voice reply back in 5 seconds
- Works both ways — RU→EN and EN→RU
Results
Numbers that matter
<5 sec
Full voice-to-voice cycle
0
Tool-switching required
Both
RU ↔ EN directions
Tech stack
PythonWhisper APIGPT-4oElevenLabs TTSTelegram Bot API