Course Objective & Scope
The voice channel brings engineering challenges completely different from text chat: background noise, misrecognised brand or product names, mispronounced numbers and abbreviations, and sub-second latency expectations. This course teaches the design and optimisation of enterprise voice AI pipelines (STT → LLM → TTS).
Participants work hands-on with Turkish and multilingual speech recognition (Whisper, Deepgram), natural speech synthesis (ElevenLabs, Cartesia, Azure/Google TTS), custom vocabulary boosting and SIP/WebRTC audio streams to deliver a working voice application and a latency/quality measurement plan.
Who Should Attend?
- Software and AI engineers building voicebots, conversational IVRs and call analytics
- Contact centre infrastructure and voice integration (SIP/WebRTC) specialists
- Technical teams developing voice product experiences
- Prerequisite: Python or Node.js and basic familiarity with APIs and digital audio
Key Takeaways
- Build an end-to-end voice pipeline combining streaming STT, LLM and streaming TTS
- Optimise STT accuracy for brand names, alphanumeric codes and Turkish morphology
- Deliver natural prosody, pronunciation lexicons and sentence-chunked streaming in TTS
- Integrate audio streams with telephony and media frameworks (SIP/PSTN, Twilio, LiveKit, Pipecat)
- Measure Word Error Rate (WER), speech naturalness and end-to-end latency
Curriculum
Voice AI Architecture and Audio Fundamentals
- Cascaded (STT → LLM → TTS) pipelines versus native speech models
- Sample rates (8kHz telephony versus 16/24kHz web), codecs and noise suppression
Speech-to-Text (STT) and Turkish Optimisation
- Streaming and batch usage of Whisper, Deepgram and cloud STT engines
- Keyterm boosting, entity/code recognition and speaker diarization
LLM Generation for Speech and Text Normalisation
- Generating concise, markdown-free responses suited to spoken delivery
- Normalising dates, currency amounts, IBANs, acronyms and phone numbers for speech
Text-to-Speech (TTS) and Streaming Synthesis
- Low-latency streaming synthesis with ElevenLabs, Cartesia and enterprise TTS
- Pronunciation lexicons, pacing/emotion controls and voice cloning safeguards
Telephony Integration and Quality Measurement
- Setting up call flows with Pipecat / LiveKit and SIP trunks
- WER, MOS and stage-by-stage latency measurement plan
Other Courses in This Category
-
Real-time AI assistants that interact instantaneously over voice, camera and screen sharing: full-duplex streaming, barge-in handling, live visual context, in-flight tool calling and resilient session management.
- Live Course
- 5 modules
-
Computer-use systems that operate desktop and browser environments via screenshot perception, coordinate grounding and keyboard/mouse actions: Claude Computer Use, OpenAI CUA, sandboxed VMs and safe execution.
- Live Course
- 5 modules
-
UI agents that execute workflows across web portals and legacy enterprise screens without APIs: DOM plus visual grounding (Set-of-Marks) with Playwright, Stagehand and Browserbase, session handling and resilient automation.
- Live Course
- 5 modules
-
User control and confirmation UX in voice, multimodal and UI-acting agents: action previews, tiered permissions, live takeover/handback, undo mechanisms and UI safety patterns.
- Live Course
- 5 modules
-
Testing voice, multimodal and UI-acting agents for user experience, trust and safety before launch: user trust calibration, adversarial scenarios, frustration signal detection and release criteria.
- Live Course
- 5 modules
-
Multimodal model architectures that unify vision, audio and text representations: ViT/SigLIP and audio encoders, cross-modal projection/fusion layers, and LoRA fine-tuning of open-weight VLMs (Qwen-VL, Llama, Pixtral).
- Live Course
- 5 modules
Get in touch
