I.N.F.E.R.N.O.S.

Interactive Neural Framework for Efficient Realtime Network Operations on Streams

🔥 Welcome to Infernos, where data comes to life in real-time! 🔥

Overview

Harness the power of I.N.F.E.R.N.O.S. to transform audio, video, and text streams with state-of-the-art inference in an instant. Embrace a blazing-fast future, free from lag.

News

Work is underway to scale TTS performance 50x. Stay tuned.

I am going to present at FOSDEM 2024 (RTC track). See you all there!

Features

• Interactive: Infernos isn't just another tool; it's an experience. Speak in one voice and marvel as it's automatically translated into a completely different tone or even language, and then seamlessly transmitted in real-time during phone or web meetings.

• Neural Power: With deep learning at its core, Infernos is optimized for top-notch performance.

• Multimodal Support: Whether it's audio, video, or text, Infernos handles them with elegance.

• Efficiency: Designed for low-latency, high-throughput operations.

• Realtime: Don't wait. Experience the magic as it unfolds.

Quick Start

1. Clone the repository:

   git clone https://github.com/sippy/Infernos.git

2. Navigate to the project directory and install dependencies:

   cd Infernos && pip install -r requirements.txt

3. Create a configuration file. In the following example we would listen and accept SIP calls from MY_IP and pass them into Live Translator application. Then use SIP account to send outbound call legs to DEST_NUM@MY_SIP_SRV:

   MY_IP="A.B.C.D"
   MY_SIP_SRV="E.F.G.H"
   DEST_NUM="12345"
   DEST_USER="foo"
   DEST_PWD="bar"
   cat > config.yaml <<EOF
   sip:
     settings:
       bind: ${MY_IP}:5060
     profiles:
       me:
         sip_server: ${MY_IP}:*
         sink: apps/live_translator/configuration1
       bar:
         sip_server: ${MY_SIP_SRV}:5060
         username: '${DEST_NUM}'
         password: '${DEST_PWD}'
         register: True
   apps:
     live_translator:
       profiles:
         configuration1:
           stt_langs: ['en', 'pt']
           tts_langs: ['pt', 'en']
           outbound: sip/bar;cld=${DEST_NUM}
   EOF

4. Light the fire:

   python Infernos.py -f -L ~/Infernos.log

5. Use SIP device or software such as Linphone to place a SIP call to sip:anything@localhost:5060. Replace localhost with a local IP of machine running Infernos if testing over LAN.

Ready to experience real-time inferencing?

Contribute

Feeling the warmth? 🔥 Eager to stoke the flames of Infernos? Delve into our contribution guidelines and join the firestorm!

License & Acknowledgements

Powered by the 2-clause BSD license. A heartfelt shoutout to the community for their priceless insights and tireless contributions.

Media

• Setting up live translation service with Infernos: Live presentation during OpenSIPS Summit 2024 setting up realtime in-call translation inference service for Portugese->English / English->Portugese on a AWS instance "from zero to hero" in less than 60 minutes.
• Infernos: cost efficient AI inference for real-time applications: Overview of the Infernos architecture and progress over the past few months.

Join US

• Discord

---

Stay on the lookout for more sizzling updates, and always remember: Infernos makes the future sizzle!