Conversations Squared #9 Andrew Muir

We’re used to hearing Text To Speech in conversational AI, but that’s not the only option available. Voice assistants don’t have to use synthetic speech - they can use recordings of a live voice talent instead. Andrew Muir, Lead Dialogue Designer at PolyAI, explains the benefits of using voice talents. One of his points really resonated with me - why voice talents ease users into the experience! 00:00 Start 00:24 Tell me about yourself? 02:38 What's your favourite bot? 03:26 What are your biggest concerns about ChatGPT? 05:58 What journalistic skills do you use in conversation design? 08:42 Why do you choose to use voice talent at PolyAI? 13:10 So voice talent eases users into the experience? 13:52 Where would you choose to use synthetic voices? 15:56 Are you using voice talents and cloned voices? 17:03 Please tell me about your relationship with the voice talents? 19:14 Do you give the voice talent freedom to create their work? 19:50 Do voice talents push back against being used in AI projects? 21:36 Voice talents could get left behind by synthetic voices 22:46 Are voice talents proactively cloning their voices? 23:21 How do you approach updates to avoid spiralling costs? 26:24 Do you adjust the design based on a voice actor's performance? 30:40 Is a synthetic assistant with an empathetic voice a design sweet-spot? 31:51 Do you think there's scope for non-human character design? 36:38 The way words are said matters 38:56 The user observes everything about the bot 41:16 How do you approach voice localisation at PolyAI? 45:44 Personas are received differently in different locales 46:43 You have to know the culture intimately to localise 48:47 We have to expect we will keep updating our work 50:35 Where can people find out more about you? Hosted on Acast. See acast.com/privacy for more information.