Text-to-speech prompting tips @AssemblyAI
Text-to-speech prompting tips  @AssemblyAI
Uploaded October 2026 | Updated November 2026, 3 hours ago
How to prompt a text-to-speech model for voice agents: hyphens for stutters, ellipses for pauses, capitals and punctuation for emphasis, and how AssemblyAI TTS reads currency, email addresses, dates, abbreviations and phone numbers. Every tip is tested live in the AssemblyAI playground.

The full prompting guide, as a markdown file you can hand to your agent: assemblyai.com/docs/tts/prompting-guide.md?utm_medium=social&utm_campaign=tts-prompting-tips&utm_source=youtube

Try our API for free → assemblyai.com/dashboard/signup?utm_medium=social&utm_campaign=tts-prompting-tips&utm_source=youtube

Pacing and tone come from how you write the text. Hyphens make a voice stutter over a word, three dots add a pause (where you put them matters), and capitals, exclamation marks and questions change what the voice emphasizes and how a sentence ends. Filler words are written out, and two letters are enough: "hm" comes out as "hmm". A word followed by a hyphen and a space gives you a false start.

Voice agents spend most of their time reading out numbers, addresses and confirmation codes, so the second half covers normalization: a dollar sign makes an amount read as dollars, email addresses are read out as addresses, a date written in numbers is read month first (even 24/10/2026 came out as October 24th), "St" is read as "Street", and common abbreviations like USA and MIT are read as expected. For a less common acronym, put hyphens between the letters. And for numbers, hyphens group the digits, while a comma makes "4,321" read as a whole number.

Chapters
0:00 Prompting tips for text-to-speech
0:36 Pacing: stutters and pauses
1:34 Emphasis and tone: capitals and punctuation
2:19 Filler words
2:45 False starts
3:02 Normalization: currency and email addresses
3:48 Dates, times and street addresses
4:59 Abbreviations and acronyms
5:33 Phone numbers and number formatting
6:28 Recap

▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT ▬▬▬▬▬▬▬▬▬▬▬▬

🖥️ Website: assemblyai.com
🐦 Twitter: twitter.com/AssemblyAI
🦾 Discord: discord.gg/Cd8MyVJAXd
▶️ Subscribe: youtube.com/c/AssemblyAI?sub_confirmation=1
🔥 We're hiring! Check our open roles: assemblyai.com/careers

▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬

#MachineLearning #DeepLearning
Text-to-speech prompting tipsBuild a Chatbot with Claude 3.5 Sonnet and Audio Data (in Python)Universal‑3.5 Pro is now available on OpenRouter with the same API key!Build an AI Voice Translator: Keep Your Voice in Any Language! (Python + Gradio Tutorial)We drove an ice cream truck around SF Tech Week | AssemblyAIVoice AI: Beyond Transcription with Granola, CoLoop & EdgeTierWe Turned Old Talking Toys Into Voice Agents | AssemblyAI SF HackathonWhat Actually Defines Success for Voice AI Agents? 🤖Universal-3 Pro Streaming Speaker Labels demoDovetail Customer Story | AssemblyAINovember 2025 RecapThe Fundamentals of LLM Text Generation
AssemblyAI |

Text-to-speech prompting tips

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER