Slides: Optimizing inference for voice models in production - Philip Kiely, Baseten

Source Video

Optimizing inference for voice models in production - Philip Kiely, Baseten

Relationship To World's Fair 2026

These slides are extracted from a public AI Engineer YouTube video connected to World's Fair 2026. Speaker-matched clips are supporting context unless later confirmed as exact session recordings; official livestream recordings are day-level/event-level source material.

Related Scheduled Sessions

• No individual scheduled session mapping has been assigned yet; treat this as an event livestream deck.

Extracted Slides

OCR text:

aWS

ee)

@®Graphite WW Windsurf MongobB

#Mdaily £3augmentcode WorkOS

OCR text:

Hi, I'm Philip from Baseten 7

« Developer relations at Baseten a te a a ° — . :

+ Based in SFBA @@@ ai a a 2a

* Favorite voice model: Orpheus TTS 4 ia | : »

od woe

- 3

Te ; ; oO

a Microsoft OCU

OCR text:

Agenda |

1. TTS model architecture

2. TTS performance metrics

3. Orpheus TTS optimization techniques

4. Orpheus TTS performance benchmarks

5. Infrastructure and client code

a 3

u Microsoft § ou?

, os

OCR text:

Example: e Llama 3.2 3B backbone

Orpheus TTS e Increased vocab size for

speech-specific tokens

V Canopy Labs —.. Extended context ienaih

with RoPE scaling

= 3

2“ " a Microsoft simon

OCR text:

¢ Optimize for Hopper architecture

TensorRT-LLM ° Post-training quantization to FP8

cottings ¢ Quantize KV cache

| - 3

a Microsoft @yr{?

OCR text:

Al Engineer

World's Fair

Slide-Derived Subjects To Review

Subject extraction uses video title, related session titles/descriptions, transcript context, and OCR text when available. OCR is best-effort and should be reviewed against the embedded slide images.