In this episode, we speak to Laetitia Cailleteau, Managing Director | DATA & AI Europe Lead | Global Lead for Conversational AI at Accenture, about the origins of the patent, the problem it solves, the value of personalised voice experiences and more about what Accenture is up to in AI, and how its thinking about conversational AI's potential to transfer customer experience.LinksConnect with Laetitia Cailleteau on LinkedInMore about Accenture See acast.com/privacy for privacy and opt-out information.

VUX World

Hyper-personalised voice AI with Laetitia Cailleteau, Accenture

We take it for granted that, when we talk to our voice assistants, they hear us. When we dictate a message to Siri, it transcribes it. Speech recognition is used in a whole manner of areas beyond voice assistants, too. Transcribing videos on YouTube, PowerPoint, dictation, call transcriptions in contact centres and beyond. But speech recognition systems don't always perform as expected and require training constantly to address bias, new language and dialects.Why do automatic speech recognition systems need training?There are many cases, however, where speech recognition systems either fail, or need training. For example, training speech recognition models to remove bias is a core priority for all speech recognition providers today. Studies have shownthat the major ASR players technologies perform poorly when used by black people, as compared to white people, for example. Addressing this performance gap is of upmost importance.Another reason for training ASR systems is expansion into other countries and to cater to other languages and accents. When Amazon, for example, aims to launch Alexa in, say, Brazil, it needs to train its speech recognition models in the Brazilian language, dialect and accents. Understanding languages, accents and dialects is a core part of making AI accessible to the world, as well as for commercial expansion.Finally, companies need to continue training ASR systems to make sure that new vocabularies and phrases that enter a given language are understood. Prior to 2020, the world wasn't saying the phrases 'COVID', 'COVID 19', 'SARS COV 2' half as much. A few year's prior, Tik Tok wasn't a thing. Language changes, and so ASR systems need to change with it.In this episodeIn this episode, we chat to Esteban Gorupicz, CEO, and Alejandro Heredia, Technology Business Development and Sales, at Atexto about how they help ASR companies train speech recognition systems and tackle the above problems.About AtextoAtexto is one of the world's leading providers in speech recognition training. It's platform allows ASR providers to crowd source training data for speech recognition systems, label and tag the data, then export it to feed into its models. It also has an ASR benchmarking tool, which allows providers to compare performance across ASR systems.LinksCheck out AtextoFind out about the ASR World Championships  See acast.com/privacy for privacy and opt-out information.

How to train an ASR (automatic speech recognition) engine with Esteban Gorupicz and Alejandro Heredia, Atexto

Swisscom, Switzerland's largest telco, has one of the most widespread uses of conversational AI we've seen to date. With voice-enabled set-top-boxes, customer service-based chat and messaging applications, as well as a voice-enabled IVR system for call routing, Swisscom is all-in on voice and conversational AI. Roger Dill heads up the Dialogue Management, AI and ML product and has scaled its efforts from a single chatbot, to a sophisticated multichannel suite of conversational capabilities. Artificial Solutions, the company behind Microsoft-LUIS-powered voice development platform, Teneo, has seen its fair share of enterprise conversational AI deployments. CEO Per Ottosson, formerly IPSoft Amelia, is well versed in the conversational AI space and, given that he's been involved in countless enterprise voice and conversational AI rollouts, including that of Swisscom, has unique perspectives into the trends, strategies and direction of the market. Hear from both Per and Roger about what it takes to scale conversational AI across the business. We cover areas such as:Organisational skills and team requirementsQuantifying and measuring success and the challenges surrounding thatSwisscom's approach to its technology architecture and its modular approachArtificial Solutions' approach to tooling and why developers are key to successful conversational AI deploymentsChallenges you'll face when scaling solutionsConsiderations for those getting startedAnd much, much, much moreLinksConnect with Per Ottosson and Roger Dill on LinkedinTry Teneo  See acast.com/privacy for privacy and opt-out information.

Scaling conversational AI with Roger Dill, Swisscom, and Per Ottosson, Artificial Solutions

Sam Danby of Boost.ai joins us to share how to take your conversational AI to the next level using Boost technology, including creating conversational agents from web content, assessing and choosing the right technology such as ASR and NLU, as well as learning and iterating on language models.LinksFollow Sam on Twitter and connect on LinkedInVisit Boost.ai See acast.com/privacy for privacy and opt-out information.

Giving your AI a Boost with Sam Danby

Marie Kleinert and Ray-Allen Taylor of VUI Agency join us to share some of the methodologies and practices they use to create charismatic voice assistants and how you can do the same.We dive into areas such as:Defining charisma from a voice assistant perspectiveTTS tuning and challengesThe value of creating a personaThe importance of linguistics in conversation designWhen and how to create charismaHow charisma contributes to improving customer experienceLinksCheck out VUI.agency  See acast.com/privacy for privacy and opt-out information.

Designing charismatic voice assistants with Marie Kleinert and Ray-Allen Taylor, VUI Agency

Google Cloud's Head of Conversational AI and Contact Center, Antony Passemard, joins us to dive under the hood of Google CCAI; Google's Contact Centre AI product.Antony manages the entire product suite and joins us to share what CCAI is, how it's being used to improve customer experience within contact centres across industries, as well as describing how you can use the tools yourself to streamline service delivery using conversational AI.Running order:00:00:00 Intro00:02:53 About Antony Passemard00:04:16 Does CCAI have a relationship with Google Assistant?00:06:35 Overview of CCAI00:07:20 What drove the need for DialogFlow CX?00:10:42 Can you move from ES to CX?00:11:50 Who is CX for?00:13:00 Core reasons for using DialogFlow CX vs ES00:17:30 Michael Kenny: Will DF CX open up third part integrations for CMS etc?00:19:40 What has CCAI taken from Google Research?00:23:13 BERT: what's it being used for?00:26:10 What is Agent Assist?00:28:50 What does Agent Assist look like and where is it?00:31:04 How is Agent Assist implemented?00:35:00 How will Agent Assist improve over time?00:37:05 How do you approach feature building and roll out?00:41:00 Focus on successful conversations00:42:50 Can you implement Agent Assist on your own?00:47:00 How to implement DialogFlow into contact centres00:52:35 Insights00:57:00 Can you use Insights to find and prioritise use cases?00:58:35 How to measure successful conversations01:05:05 How to find out moreLinksFind out more about Google CCAI  See acast.com/privacy for privacy and opt-out information.

Google CCAI with Antony Passemard

Voicebots for customer service are in increasing demand, but how do you actually integrate one into a contact centre?Learn how to take an existing chatbot and repurpose it for your contact centre. Many chatbots provide immense value to customers, and many call centres are overwhelmed with demand. Why not offer your chatbot service through your contact centre and help customers self serve, and customer service staff manage demand.Joining us this week is Ilan Avner, Director of Product Management at AudioCodes. Ilan will share how any brand or developer who has existing conversational AI implementations can scale those conversations into ANY contact centre system.We discuss the challenges you'll face when trying to integrate bots into contact centres including how to deal with on-premise vs cloud deployments, security, protocols and standards, timescales and costs. We'll also cover how you can avoid all of the completely by using the AudioCodes Voice AI Connect product.A truly enlightening conversation for anyone who's interested in the practicalities of integrating any kind of bot into contact centres for improving self service and customer experience.LinksCheck out the AudioCodes Voice AI ConnectCatch up on the AudioCodes webinar with VodafoneConnect with Ilan Avner on LinkedIn See acast.com/privacy for privacy and opt-out information.

Integrating ANY bot from ANY framework into ANY contact centre with Ilan Avner, AudioCodes

As much as voice technology has been hyped over the last few years, perhaps leading to some unrealistic expectations, it's beginning to find its feet in reliably and consistently improving customer service in contact centres. Regardless of hype, voice technology has super-human qualities when applied in a customer services environment and especially, according to today's guest Nikola Mrkšić, CEO and co-founder of London-based PolyAI, when using PolyAI technology.We chat to Nikola about what defines 'super-human' qualities for voice assistants, how the PolyAI technology differs from some of the common providers, why drag and drop tools have limitations, and understand Nikola's perspective on the voice industry overall.LinksVisit PolyAIFollow Nikola Mrkšić on Twitter or LinkedInWebinar on multilingual voice assistants See acast.com/privacy for privacy and opt-out information.

Creating super-human voice assistants with Nikola Mrkšić, CEO, PolyAI

Tom Hewitson, Founder and CEO, Labworks.io, joins us to share how he built a business on Alexa and Google Assistant, how he finds, keeps hold and monetises users, design best practice and advice for those looking to do the same.In 2017 and 2018, when Amazon Alexa began rolling our in-skill purchasing (ISP), many thought that there'd be a gold rush on the Alexa platform. Business will be built, money will be made and all will be well. It happened with the App Store on mobile, after all.There was two problem that most developers were facing: how do you create something that's good enough to keep users coming back for more, and then how do you create an offer for those users that's worth paying for?Those challenges turned out to be pretty difficult to overcome and, to date, only three companies have managed to reach the levels of success many were aiming for in the Alexa space: Matchbox, with their suite of quizzes and kids games, Invoked apps with it's load of sleep and ambient sounds and, our guest this week, Labworks, with Voice Arcade.One weekend, wanting to explore Alexa for investigatory purposes, in case any clients were to ask about it in future, Tom Hewiston, founder of then-chatbot agency, Labworks, set about building a game.After launching the game and pretty much forgetting about it, some months later, Tom received a call from Amazon. Turns out, Tom's game was the third most popular game on the Alexa platform. Amazon were starting to pay developers for having popular skills as an incentive for them to continue improving skills and developing new skills.These rewards were enough for Tom to employ someone to help build out more games. Labworks pivoted from building chatbots for clients and into building games for Alexa and the rest is history.We chat to Tom about his journey, what he's learned, his thoughts on the industry and advice for others looking to build successful skills and actions on Alexa and Google Assistant.Here's the running order:00:02:50​ About Tom Hewitson and Labworks.io00:05:50​ Last year's Alexa skills usage00:09:30​ Was it good timing to launch Voice Arcade?00:11:15​ Ten hot takes on voice: what's changed in voice since 2019?00:13:00​ What's changed for the better?00:15:20​ Developer rewards: how helpful were they?00:19:15​ What works and what doesn't for in-skill purchasing?00:20:00​ What's the difference in typical ISPs vs Voice Arcade?00:22:15​ Challenges with ISPs, offering more value than you're asking for in return00:24:30​ How are you showing value?00:26:00​ Vishnu Saran: How are you solving discoverability?00:30:40​ Marek Mis: Advice for hobbyist voice devs wanting to go pro?00:33:40​ Is there much experimentation left?00:38:40​ How do you prioritise and test features?00:40:55​ Any new platform features you're excited about?00:43:00​ Matt Buck: Have you thought about putting your games in apps or on web?00:47:09​ Katy Boungard: Brand identity and sonic branding00:49:20​ Richard Warzecha: Discovering features within skills00:50:00​ How much time is spent on happy path vs everything else?00:55:44​ Manual testing vs code testing00:57:05​ Example of something that took a lot of iteration to get right00:58:35: Richard Warzecha: Transferable principles from games to other voice experiences01:03:00​ Tooling and Jovo01:04:00​ General takeaways and contact detailsLinksLabworks.ioFollow Tom Hewitson on TwitterJoin Tom Hewitson on LinkedInSay "Alexa, open Voice Arcade"Say "Hey Google, open Voice Arcade" See acast.com/privacy for privacy and opt-out information.

Building a business on Alexa and Google Assistant with Tom Hewitson

Derek Roberti is the VP Technology, North America, at Cognigy, based in California. He joins us to share his insights and observations on using conversational AI automation in the enterprise for automating service delivery. We chat to Derek about:Cognigy and its historyDemocratising AIThe need for market education The balance between technical capability and ease of use in conversational AI toolsMarrying conversational AI with Robotic Process Automation (RPA)Whether conversational AI tools will convergeHow and if we'll see conversational automation within voice assistantsLinksVisit Cognigy.comConnect with Derek Roberti on Linkedin See acast.com/privacy for privacy and opt-out information.

Enterprise conversational automation with Derek Roberti

Nate Brown is the Chief Experience Officer at Officium Labs, world-leading CX consultancy. He joins us to share how a customer-centric culture shaped around customer feedback can create the conditions and insights needed to build products and services that customers love. The exact same methodologies and practices can feed directly into your conversational automation strategy to make sure you're consistently meeting customer needs.At the same time, the data and insights you produce from the automated conversations you have with customers can be like gold dust, and can be fed back into the voice of the customer operation in the rest of the business.We chat with Nate about:The value of a voice of the customer strategyThe importance of staff feedbackVoice of the customer processThe listening pathEmpowerment and actionSentiment vs intentImplementation challengesLinksOfficium LabsConnect with Nate Brown on LinkedIn See acast.com/privacy for privacy and opt-out information.

How voice of the customer can enhance your conversational automation strategy with Nate Brown

When it comes to making customer interactions with conversational AI more human-like, there's often one thing missing. Emotion.We chat to Nico Perony about how OTO.ai is being used to understand emotion and sentiment of users in order for businesses to create more compelling customer experiences.Married with conversational AI, bots can be more sensitive to the caller based on their emotional state and even do things like escalate calls to an agent based on an angry customer response.It's also being used to support agents in call centres to appraise tone and improve how agents speak with customers, as well as in gaming applications to detect abuse and toxicity and escalate occurrences to moderators.There is huge potential for this technology, not only in automating customer experience in a more human-like way, but also to gather insights from customer and staff voice relations and use it to improve the quality of service offered.We also continued this discussion on Clubhouse afterwards. Follow @kanesimms on Clubhouse for more open ended after show discussions.LinksVisit OTO.aiConnect with Nico on LinkedIn See acast.com/privacy for privacy and opt-out information.

Actionable insights from speech data with Nico Perony

When I was on the Voicebot podcast, Bret asked me whether we're going through a voice first winter. I said that I didn't think so, nor did I really care. There are opportunities to do good work and enhance people's lives regardless. However, there has been an undercurrent in pockets of the industry where some folks are becoming frustrated that voice hasn't reached its potential (however you define that).Over the past few weeks, founder of WitLingo, Ahmed Bouzid was involved in a conversation at the popular VoiceLunchmeet where a participant stated that:‘Beyond the weather, time, and the occasional timer and alarm,’ they mused out loud, ‘am I myself really using my Amazon Alexa and Google Assistant that much in my life? I mean, really and honestly, am I? No, actually, not really…. So, if I am not using them that much and yet I am such a believer in Voice First, what hope is there for the rest of the world?’He wrote a piece summarising his thoughts. This coincided with a post from Bespoken.io founder, John Kelvie's post claiming voice is stuck.The primary questions raised in both of these two pieces are:Is voice living up to its promise?Where is the true value in voice first?Has there been enough improvement in the technology and ecosystem to support its own ecosystem like the App Store?To debate this, we brought along Voicebot.ai founder Bret Kinsella, together with Ahmed and John to figure out where we actually stand and where we need to go next. We also continued this discussion on Clubhouse afterwards. Follow @kanesimms on Clubhouse for more open ended after show discussions.LinksRead voicebot.aiCheck out WitLingo and BespokenRead Voice is stuckRead The #VoiceFirst User Interface Has a Use Case Fit ProblemRead Voice first sucksFollow Ahmed, John and Bret on TwitterConnect with Ahmed, John and Bret on LinkedIn See acast.com/privacy for privacy and opt-out information.

Does voice first suck? And is it stuck? With Bret Kinsella, John Kelvie and Ahmed Bouzid

Resemble.ai is a cutting edge synthetic voice (text-to-speech) provider that allows anyone or any brand to create their own, customer TTS solution to use across any and all channels. From IVR, Alexa and Google Assistant to narrating videos and marketing materials and even voicing game characters. With just 50 lines of speech, you could have your own dedicated brand voice that brings a consistent and elevated customer experience to your conversational applications. With quicker-than-real-time processing, the low latency capabilities makes having a conversation with a Resemble.ai-voiced assistant as natural as can be.In this episode of VUX World, we're joined by founder and CEO, Zohaib Ahmed, to discuss the work Resemble is doing to help brands create stand-out conversational experiences. We discuss what Resemble is and how it works, the value of having your own custom voice, the use cases and applications it's being used for, the process for creating, the editing and publishing tools for customising intonation and emotion, as well as the ethical considerations around voice cloning, fraud and deep fakes.00:00 Intro02:55 About Zohaib Ahmed and Resemble.ai04:47 What are people using Resemble.ai for today?07:37 How much dialogue is needed to create a TTS voice?08:27 Voice Talent Pool: what's that?11:50 Bondad: How far off are we from high quality rendering at run time with low latency?14:00 What's the process for creating a custom brand voice?17:00 Are call centres cloning staff voices?19:00 An alternative approach from a branded assistant21:10 Ethics of voice cloning24:20 Resemblyzer and tackling speaker verification27:40 Matthew James-Dewstowe: Can you tweak the voice once its created?31:40 Jay Ruparel: How do you handle dynamic speech and keep the voice integrity?37:40 Sean Thornton: Predicting emotion41:00 Heidi Cohen: Regionalisation and localisation44:45 Real time translation of speech48:20 Jay Ruparel: Thoughts on filler words like 'um', 'ah' etc51:37 Proving the value and ROI01:00:40 Where can people find out more about Resemble.aiLinksVisit Resemble.ai Follow Resemble.ai on TwitterZohaib Ahmed on Twitter and LinkedIn See acast.com/privacy for privacy and opt-out information.

Bringing human-like performance to AI custom voices with Zohaib Ahmed

Mihai Antonescu joins us to discuss the history of voice at Mercedes, the origins of the HeyMercedes voice assistant and the process for building it.You'll also learn tips and tricks that'll help you put together your approach to creating your own assistant, as well as how to understand and measure the business value voice assistants can bring.Mihai asked us before the show to challenge him with our questions, and we certainly hope we did.LinksConnect with Mihai Antonescu on LinkedIn or Twitter See acast.com/privacy for privacy and opt-out information.

How we built the HeyMercedes Voice Assistant with Mihai Antonescu

Learn how to implement voice first strategy and create world-class voice experiences with VUX World. Every week, we speak to industry thought leaders and practitioners on the three core pillars of voice: strategy, design and development. We share the deep details of how they do what they do, so that you can do what you do, better.