July 2026 Summaries

Multilingual transcription involves converting spoken audio into written text across multiple languages, utilizing advanced Speech to Text (STT) models that incorporate features like automatic language detection, speaker diarization, keyterm prompting, and entity detection. These tools are crucial for various industries such as telecom, finance, and healthcare, as they enhance accessibility, searchability, and compliance by creating accurate, searchable records from audio data. ElevenLabs’ Scribe v2 model exemplifies these capabilities, supporting over 90 languages and offering features like real-time transcription and high precision with low Word Error Rates for many languages. The pricing for these services typically depends on audio duration rather than the number of languages, and they can be integrated into production workflows via APIs, allowing organizations to efficiently manage multilingual transcription at scale.

Creating a full-cast audiobook has become more accessible with ElevenCreative, which simplifies the traditionally complex process into five straightforward steps: uploading the manuscript, casting characters, setting pronunciations, directing pacing, and generating the final product. Unlike traditional methods that require extensive coordination and high costs, ElevenCreative's innovative Character Casting feature automatically detects characters and assigns unique voices, eliminating the need for a recording studio or professional experience. This approach significantly reduces production time and cost, making it an attractive option for indie authors and smaller publishing studios. While traditional narration may still hold value for some projects, ElevenCreative's platform offers a practical solution for authors looking to self-publish with a fraction of the effort and expense.

Character Casting in Audiobooks is a cutting-edge tool designed to bridge the gap between the voices authors envision for their characters and what listeners hear, by detecting characters in a manuscript and proposing suitable voices for each. Authors can upload their manuscripts to have characters automatically identified, with voices previewed using real dialogue from the book, allowing for customization until the desired voice is achieved. The platform also includes features like a pronunciation dictionary for unique names and terms, and configurable pacing for paragraphs and chapters. With support for narration in over 90 languages and a library of more than 10,000 voices, including the option to clone one's own voice for narration, Character Casting significantly reduces the time and cost of traditional audiobook production, making it an attractive option for authors, self-publishers, and professional publishers aiming to reach a wider audience through audiobooks.

Valiant Finance, an Australian company connecting small and medium-sized businesses with over 90 lending partners, has integrated ElevenLabs' voice solutions to enhance its customer interactions and streamline its lending process. By employing tools like ElevenAgents and ElevenCreative, Valiant has improved after-hours service with a voice agent named Sophia, who handles 29% of after-hours calls, ensuring no customer inquiries are lost overnight. The marketing team has also expedited the production of radio campaigns, creating 10 ad variants in one week, significantly reducing costs and time. With an ongoing partnership with ElevenLabs, Valiant aims to reduce administrative tasks and enhance broker training through internal simulation tools, while also developing an in-hours agent to match previously unsupported customers with new financing options as their lending panel continues to expand.

AI virtual receptionists are software systems designed to handle incoming calls for businesses, understanding and responding to callers' needs using natural language processing. Operating around the clock, these systems ensure businesses never miss a call, providing capabilities such as call routing, appointment management, and lead capture. They integrate seamlessly with various tools like CRMs, calendars, and telephony systems to autonomously complete tasks such as booking appointments and updating records. AI receptionists offer advantages like 24/7 availability, consistency in customer interactions, scalability during call spikes, and multilingual support, making them particularly beneficial for small businesses that may not have the resources for full-time front desk staff. By reducing wait times and providing immediate responses, AI receptionists enhance customer experience and improve operational efficiency while offering a substantial return on investment compared to traditional receptionists.

ElevenAgents offers a versatile platform for deploying customer service agents across multiple communication channels, including SMS, Telegram, Intercom, and Freshdesk, alongside traditional platforms like phone, web, Zendesk, Slack, and WhatsApp. It enables businesses to define their agent's knowledge and procedures once and deploy them universally, ensuring a consistent customer experience. The platform's redesigned channels dashboard simplifies the management and integration of these channels, allowing users to see live connections and add new ones effortlessly. ElevenAgents also provides tools for tuning agent behavior per channel and conducting simulations to ensure quality before updates go live, thereby enhancing customer interactions. The platform continues to expand its capabilities and channel integrations, currently in Alpha for some, to include diverse modalities like handling images, files, and audio.

Medical speech-to-text (STT) technology transforms spoken clinical interactions into accurate, structured documentation, addressing the challenges of manual data entry and memory reliance faced by healthcare professionals. This technology integrates speech recognition with clinical terminology detection, enabling real-time transcription that maintains high accuracy despite accents, noise, and complex medical vocabulary. Key features of effective medical STT software include low latency, speaker diarization, multilingual support, and compliance with healthcare regulations such as HIPAA. These systems seamlessly integrate into existing electronic health record (EHR) workflows, improving documentation efficiency and reducing the administrative burden on medical staff. Real-world applications span various healthcare settings, from outpatient clinics to telehealth consultations, where they enhance the speed and accuracy of record-keeping, ultimately allowing healthcare providers to focus more on patient care.

The text discusses the role of voice AI technology in enhancing the democratic process by improving access to electoral information and combating misinformation, with a particular focus on the upcoming 2026 elections. Voice AI provides essential services for visually impaired individuals, those with low literacy, and non-native speakers by converting complex written information into spoken dialogue. In Brazil, the Superior Electoral Court (TSE) is integrating voice AI into its 'Electoral Assistant' to facilitate voter access to official election information, and partnerships with organizations like Comprova aim to verify information and prevent misinformation. In the United States, collaborations with policymakers and tech companies focus on safeguarding elections from voice-enabled scams and misinformation. To prevent misuse of voice AI, ElevenLabs has implemented comprehensive safeguards, including "No Go Voices" and Voice CAPTCHA, and offers tools like the AI Speech Classifier to detect AI-generated content. The company collaborates with entities like Google DeepMind to enhance traceability and transparency of AI-generated audio, emphasizing its commitment to ethical use and continuous improvement of voice AI technology worldwide.

MCK Leżajsk, the oldest and largest cultural institution in southeastern Poland, leverages the ElevenLabs Impact Program to enhance accessibility and engagement in its diverse cultural offerings. Founded in the 1970s, the center serves as a hub for local culture, history, and art, running facilities like the town's public library, cinema, and historical archive while organizing a variety of community programs and events. By integrating ElevenLabs-generated audio descriptions into exhibitions and promotional materials, MCK Leżajsk has significantly improved access for individuals with visual or cognitive impairments, allowing them to experience art and events in new ways. This innovative use of AI voices extends to city-wide announcements and tourist guides, enhancing the reach and impact of the center's initiatives. The success of these efforts was evident during major events like the Great Orchestra of Christmas Charity and the Leżajsk Christmas Market, which drew record-breaking attendance. Looking ahead, MCK Leżajsk plans to further incorporate audio elements into its services, including multilingual tourist guides and AI-narrated audiobooks, to preserve and promote regional literature and culture.

Conversational AI design has evolved from static decision trees to dynamic, real-time interactions, where AI agents can draw from knowledge bases and remember previous conversations, requiring a higher level of design sophistication to maintain natural interactions. This involves configuring the AI's persona, voice, workflow, and decision points to ensure seamless, human-like interactions across different channels, with voice presenting the greatest challenge due to its sensitivity to latency and timing. ElevenAgents provides tools to optimize these elements, offering features like a vast voice library, real-time turn-taking models, and workflow design capabilities to enhance user experience and trust. Effective conversational AI design directly influences customer satisfaction, resolution rates, and reduces the need for human intervention, as evidenced by eDreams ODIGEO's successful deployment of AI agents that significantly improved resolution speed and reduced call transfers. The platform allows for nuanced customization and integration with existing systems to handle tasks like identity verification, ensuring that AI agents are not just functionally competent but also aligned with brand expectations and user needs.

References is a feature available on ElevenMusic, ElevenCreative, and via ElevenAPI that allows users to upload an existing track to guide the creation of a new music piece that matches the original's style, instrumentation, and feel. Users can combine a reference track with a text prompt for more specific creative direction, or rely solely on the reference. The feature supports uploads ranging from 10 seconds to 5 minutes and is designed to ensure copyright compliance by checking reference tracks for any ownership issues before use. It is intended for musicians who want to shape a sound they envision, teams looking to maintain brand sonic identity, and developers integrating precise music generation into their products.

Deploying conversational AI at an enterprise level presents unique challenges in compliance, integration, and scalability that smaller organizations may not encounter. Enterprise conversational AI systems are designed to automate high-volume tasks across customer support, internal operations, and multilingual environments, delivering significant improvements such as reduced resolution times and higher conversion rates. For instance, companies like Klarna and Everlywell have successfully integrated AI to enhance customer interactions, resulting in faster service and increased engagement. Systems like ElevenAgents are tailored for large-scale deployment, offering flexibility with large language models and comprehensive integration capabilities. They emphasize the importance of planning for compliance, data governance, and infrastructure integration from the outset to ensure successful deployment. Enterprises adopt these AI solutions to manage repetitive and costly customer interactions more efficiently, aiming to improve customer experiences, reduce operational loads, and maintain high service standards across diverse regions and languages.

ElevenMusic offers an innovative platform for generating original music with a consistent vocal and stylistic identity through its Vocals and Styles features. Users can utilize the Voice-to-Song tool to transform a single voice recording into a complete track in seconds, either using their own voice or selecting from a Vocal Library. The platform allows for the creation of reusable Vocals by uploading existing music, which Music v2 then learns to maintain the user's vocal tone and style across future tracks. Similarly, Styles bring consistency to the sound by enabling users to upload their own tracks or choose from the Style Library, allowing for the generation of full tracks with a cohesive auditory identity. All uploaded content is screened for copyright compliance, ensuring the protection of artists' rights. Vocals and Styles can be accessed on ElevenMusic, where users can explore and experiment with these features by creating a free account.

Rosebud AI, a platform for creating video games, has significantly enhanced its offerings by integrating Eleven Music and Sound Effects APIs, enabling automatic audio generation for over two million games. Previously, game creators on Rosebud had to manually source and integrate audio, navigating complex licensing issues. However, the new integration allows for seamless audio production, automatically generating soundtracks and effects during the game's creation process, and offering further customization options through user prompts. This development has increased creator engagement by 23% and improved the quality of games, making them more polished and shareable, with 1,500 audio-equipped games now produced daily. The integration has also increased daily active users by 8%, as more refined games become available on the platform.

Accessing high-quality YouTube video transcripts has become increasingly convenient, offering options such as YouTube's built-in transcription feature, AI transcription tools, and manual transcription. YouTube's in-app tool provides immediate, free transcripts for videos with captions but lacks download functionality and struggles with accents or low-quality audio. AI transcription tools, like ElevenCreative, offer fast, accurate transcriptions with features like speaker labels and are suitable for content repurposing, accessibility, and searchability. Manual transcription remains an option for those preferring customization but is time-consuming. Transcription enhances accessibility for hearing-impaired users and facilitates content creation, such as articles or guides, by providing a searchable text script. Each method's suitability depends on the desired use, whether for personal note-taking, professional content repurposing, or accessibility improvements.

ElevenLabs has collaborated with Ukraine's Ministry of Digital Transformation to integrate AI voice agents into government services, creating a more responsive and efficient state. This initiative includes ElevenAgents, an AI-powered voice agent deployed at Ukraine's State Employment Service contact center. The voice agent, which processes calls in natural Ukrainian, has successfully handled around 25% of total call volume, reducing the average response time from 30-40 seconds to 18 seconds. It remains operational during air raid alerts, ensuring continuous service. The project is part of a broader digital ecosystem called Obrii, aimed at proactively supporting citizens in finding jobs and accessing training. The AI system, compliant with public-sector standards, enhances data security while delivering more accessible and efficient public services, ultimately transforming the contact center into a remote service delivery hub.

The Finetunes Music API allows developers and companies to enable users to train the Eleven Music model on their own tracks, creating a personalized Finetune that reflects their unique sonic identity. This tool addresses the challenge artists and creators face in maintaining a consistent musical identity by learning from the original music provided and carrying its tone, texture, and style into every subsequent generation. Each Finetune remains private and is accessible only to its creator, ensuring that every track generated builds on the user's distinct sound. Additionally, the API's third-party content identification check helps protect artists' rights by flagging known songs before training begins, allowing users to confidently craft their music. The Finetunes Music API is available for developers on ElevenLabs, enabling both individual artists and companies to establish a recognizable brand sound.

Word Error Rate (WER) is a crucial metric for assessing the accuracy of Automatic Speech Recognition (ASR) tools by quantifying transcription errors through substitutions, deletions, and insertions relative to the total words in a reference transcript. A lower WER indicates higher accuracy, with less than 5% generally considered acceptable for most applications, though more precise requirements exist for fields like legal or medical transcription. Despite its utility in benchmarking ASR models, WER treats all errors equally, which may not accurately reflect contextual meaning, prompting the development of alternative metrics like Semantic Word Error Rate (SWER) that evaluate whether the intended meaning is preserved. Emerging research, such as by Artificial Analysis, highlights industry leaders like ElevenLabs’ Scribe v2, which achieved a 1.5% WER on the AA-AgentTalk dataset. While WER provides an objective comparison across ASR systems, it can exceed 100% due to insertions and may not fully capture the semantic accuracy, necessitating complementary assessments for more comprehensive evaluations.

SevenRooms, in collaboration with ElevenLabs, has developed Voice AI to enhance the hospitality experience by managing phone reservations for restaurants, thereby increasing revenue and improving guest interaction. The AI system is integrated into the SevenRooms platform, enabling it to instantly access guest profiles, check availability, and adhere to venue-specific policies, ensuring accurate and real-time bookings. Since its implementation, SevenRooms Voice AI has efficiently handled over 400,000 calls across more than 600 restaurants, capturing up to 25% more reservations, particularly during peak and after-hours. This solution not only alleviates staff workload by reducing phone handling time but also elevates the guest experience by providing swift and consistent responses, ultimately leading to more bookings and increased average spending through upsell opportunities.

Stray Sheep, a Bulgarian social enterprise, is committed to enhancing accessibility for Deaf individuals by bridging the gap between the Deaf community and the hearing majority through education and media. Despite the official recognition of Bulgarian Sign Language in 2021, the country still faces challenges in accessibility, with only a small fraction of the population able to sign. Stray Sheep addresses this by offering sign language courses led by Deaf lecturers and integrating sign language services into major consumer brands, thereby facilitating communication for Deaf customers. The organization also uses ElevenLabs Text to Speech technology to create layered videos where Deaf and hearing audiences can share the same experience, and this technology allows them to produce multilingual narrations efficiently. As part of their expansion, Stray Sheep aims to grow its digital education platform and produce accessible content for businesses and institutions, with plans to extend their impact beyond Bulgaria by combining sign language with AI. Through the ElevenLabs Impact Program, Stray Sheep and other nonprofits are supported in using AI audio to break down communication barriers in various sectors.

Voice AI systems often struggle to engage in natural human-like conversations due to their design, which typically processes interactions as discrete inputs and outputs. These systems fail to handle interruptions, pauses, and overlapping speech effectively, leading to conversations that feel mechanical and disconnected. ElevenLabs is developing advanced interaction models to address this issue by implementing a cascaded architecture that allows for real-time response and context continuity across exchanges. Their technology includes in-house Speech to Text (STT) and Text to Speech (TTS) models, speculative turn-taking, and expressive delivery modes, enabling AI agents to maintain the flow of conversation, adapt their tone, and perform parallel tasks without losing context. By optimizing each stage of the pipeline independently, ElevenLabs aims to create voice agents that behave more naturally, reducing the gap between talking to software and conversing with a human. These agents are already deployed in high-stakes environments, adhering to rigorous compliance standards, and are capable of handling complex customer interactions without human intervention.

The text provides a comprehensive guide on building a voice agent using ElevenLabs and Twilio within approximately 20 minutes. It outlines the integration of various technologies, including speech synthesis and transcription through ElevenLabs, telephony via Twilio, and language model interaction using either OpenAI or Anthropic. The architecture allows for component swapping, making it adaptable to different preferences. It describes the process of setting up a voice agent, highlighting the steps from handling phone calls with Twilio, transcribing audio using Scribe v2 Realtime, generating responses with a language model, and converting text to speech with Flash v2.5. The document emphasizes handling latency, concurrency, barge-ins, and maintaining security by managing API keys responsibly. It also suggests using ElevenAgents for a managed service alternative and details prerequisites and setup instructions for developers.

In a live workshop led by Luigi Sambuy, the focus was on creating AI agents that sound natural, emphasizing that while many teams have achieved accuracy in task completion, the challenge lies in delivering responses that resonate with human emotion and context. The discussion highlighted the importance of rhythm, flow, mirroring emotions, understanding context, and establishing a unique identity for agents to avoid sounding artificial. A demonstration showcased an AI's ability to handle a reservation change call at Rosette Restaurant, where the agent's natural interactions—such as acknowledging an anniversary, correcting discrepancies, and offering thoughtful suggestions—were driven by prompt engineering rather than advanced models. Key strategies for developing such agents include utilizing emotional tags, appropriate background noise, tone settings, pronunciation dictionaries, and voice design or cloning. Best practices suggest embedding tone directly into system prompts, allowing self-correction, and adapting emotional tags to suit the caller's state, while also considering agent eagerness and latency perceptions to enhance naturalness. Monitoring and testing agents through simulations and sentiment analysis are crucial for refining interactions and ensuring a human-like experience.

AI calling leverages artificial intelligence to handle inbound and outbound phone calls, offering a more intuitive and efficient experience compared to traditional Interactive Voice Response (IVR) systems. By integrating components like speech-to-text, large language models, text-to-speech, and telephony channels, AI voice agents can conduct natural conversations, significantly reducing operational costs and improving customer service outcomes. These systems not only handle routine calls, freeing human agents for more complex tasks, but also facilitate outbound campaigns by managing high volumes of calls simultaneously without fatigue. Companies such as Klarna and eDreams ODIEGO have demonstrated substantial improvements in resolution speed and customer satisfaction by adopting AI calling solutions like ElevenAgents, which provides tools for voice creation, conversation logic, and real-time analytics. With the ability to operate 24/7 and deliver consistent quality at scale, AI calling enhances both the efficiency and cost-effectiveness of customer service operations while ensuring compliance with regulatory standards.

Alpha Bank is collaborating with ElevenLabs to enhance customer experience through AI integration in its digital channels and call centers, developing a voice agent that offers a distinctive identity and supports Greek and English natural conversations. This innovation, built on ElevenAgents, aims to streamline customer interactions, providing fast and convenient service while connecting customers to specialized advisors as needed. Alpha Bank plans to broaden this initiative across its e-banking and mobile platforms, with the ultimate goal of delivering a seamless, intuitive customer experience that aligns with the bank's identity. Both Vassilios Psaltis, CEO of Alpha Bank, and Mati Staniszewski, Co-Founder of ElevenLabs, express enthusiasm for the partnership, noting its significance in the European banking sector.

Projekt Kalwaria, a Polish nonprofit, is using advanced technology to digitally reconstruct the Ujście Calvary, a historical hillside complex destroyed during World War II. By digitizing over 300 archival documents and using 3D modeling, the team has recreated the site's topography and several stations, winning awards for their design. These digital reconstructions are transformed into narrated video experiences using ElevenLabs Text to Speech, allowing multilingual access without the need for multiple voice actors. The narrated films, enhanced with AI-generated audio and video, are shared on YouTube, reaching a broad international audience. The project aims to further engage visitors through a mobile app and virtual reality, making the historical site accessible worldwide. This initiative highlights how AI technology can help preserve cultural heritage, enabling small volunteer teams to produce high-quality content and ensuring the survival of historical sites for future generations.

ElevenAgents Spotlight is an advanced tool designed to enhance the performance of enterprise agents by allowing teams to continuously iterate and improve their deployment through real-time analysis of every conversation. It automatically groups conversations into topics, assesses quality using plain language criteria, tracks sentiment, and detects anomalies, providing actionable insights for improvements in customer satisfaction, resolution, and conversion rates. By monitoring key metrics and offering proactive recommendations, Spotlight helps teams quickly identify issues and suggests next steps, ensuring that agents consistently evolve from good to great. The platform integrates seamlessly with existing operational tools like Datadog and Grafana, allowing agents to be part of the broader monitoring ecosystem without requiring teams to learn new systems.

Fyxer is an AI email assistant designed to help professionals manage their inboxes more efficiently by organizing emails, creating drafts in the user's tone, and writing meeting notes for platforms like Gmail, Outlook, and Zoom. To enhance the accuracy and efficiency of its meeting notetaker, Fyxer integrated ElevenLabs Scribe v2 for live transcription, which excels in speaker-attributed word error rate, crucial for Fyxer's outputs. After benchmarking various transcription providers, Fyxer selected ElevenLabs for its superior performance, achieving a 20% reduction in word error rate. This partnership resulted in a 15% relative lift in user conversion rates from trial to paid subscriptions during an A/B test with over 6,000 organizations. The integration not only improved transcription accuracy but also ensured teams could rely on precise meeting summaries and action items, freeing them from the task of correcting misattributed quotes and enhancing overall productivity.

CoopVoce, a leading mobile virtual network operator in Italy, has enhanced its customer service experience by integrating ElevenLabs' Text to Speech technology into its AI-powered virtual assistant. This integration, implemented through the ElevenAPI platform, addresses the challenge of delivering natural and conversational voice interactions, which were lacking in previous synthetic voice solutions. Since its deployment, the virtual assistant has significantly improved customer engagement, handling over 1.75 million API calls and generating over 3,729 hours of audio, with a notable 24% month-over-month increase in usage. The assistant has also achieved a 10% increase in customer willingness to engage, a 15-second increase in average conversation duration, and a 3–4% improvement in containment rate, allowing more issues to be resolved without human intervention. CoopVoce views voice as an essential component of customer support, aiming to blend automation with human specialists to manage both routine inquiries and more complex cases requiring empathy and judgment.

ElevenLabs has officially launched its business in Canada, emphasizing a significant investment in local teams and infrastructure, with plans to expand its workforce and open its first office in Toronto. The appointment of Max Lemmens as General Manager will steer the company's operations, capitalizing on his experience with companies like Revolut and Klarna. Canada, with its bilingual culture and diverse industries such as financial services, healthcare, and telecoms, presents a fertile ground for ElevenLabs' voice and audio AI technologies, which are already being utilized by prominent companies like TELUS Digital and The Globe and Mail to enhance customer interactions. Ontario's robust AI talent pool and thriving technology sector are key factors in ElevenLabs' decision to establish a presence there, aiming to support the province’s innovation ecosystem and create high-paying jobs. With Canada’s AI strategy targeting substantial growth in enterprise AI adoption by 2034, ElevenLabs aims to leverage this opportunity to revolutionize communication across various sectors.

Building a demo-quality conversational AI agent quickly can initially appear successful, but scaling it to handle real-world volume and complexity often exposes architectural flaws, particularly when a single agent is tasked with multiple responsibilities. This bottleneck results in slower decision-making, unreliable tool selection, and increased fragility due to a lack of structured boundaries. The article advocates for a multi-agent architecture, drawing parallels to organizational growth, where specialized teams with clear scopes enhance efficiency. By splitting tasks among specialized agents with distinct roles, such as data retrieval, fact-checking, and customer interaction, the system becomes more robust and predictable. However, this approach is not without challenges; context fragmentation and coordination costs must be managed carefully. The solution lies in selectively specializing agents for parallel, independent tasks while maintaining continuity where necessary, using platforms like ElevenAgents, which provide the infrastructure to support this nuanced balance, ultimately leading to more predictable and manageable systems capable of handling complex interactions effectively.

ElevenMusic offers a suite of AI-powered tools that assist musicians at various stages of the creative process, from initial idea to final production. These tools, such as Voice to Song and Loop Studio, facilitate the transformation of raw recordings and early musical concepts into polished, studio-quality tracks by addressing pitch, timing, and arrangement. Additional features like Genreshift and Unplugged allow artists to explore different musical styles by altering instrumentation and production while maintaining the core melody and lyrics. Designed to enhance creativity and efficiency, these tools are available on the ElevenMusic platform, providing musicians with innovative ways to develop and refine their work.

ElevenLabs' paid marketing team developed an internal workflow to localize ad campaigns across seven languages, significantly enhancing their advertising performance without increasing headcount. This initiative generated $3.78 million in incremental conversion value and earned a 2026 Google Ads Impact Award for measurable business results. Initially, the localization process was manual and time-consuming, leading to delays and revenue impacts. To address these challenges, ElevenLabs built internal tools that automated and accelerated the localization of search, display, and video ads, enabling their four-person team to efficiently manage campaigns across multiple languages. The streamlined workflow resulted in a 17.6% lift in conversions from non-English campaigns, demonstrating the effectiveness of their approach. This internal tool, now available as Ads Engine, connects to ad accounts, automates creative localization, and optimizes ad performance, offering a scalable solution to brands facing similar localization challenges in digital marketing.

ElevenLabs has launched Singapore data residency for its enterprise customers, enabling companies operating in or serving the region to utilize ElevenAgents, ElevenCreative, and ElevenAPI while adhering to local data residency, security, and latency requirements. This initiative allows enterprises to host data and run inference entirely within Singapore, ensuring compliance with local regulations and offering low latency for users in Singapore and East Asia. The service is exclusively available to enterprise customers, who can contact their customer success manager for implementation. ElevenLabs is committed to expanding its presence in Singapore, already collaborating with companies like Scoot, Atome, and Funding Societies to integrate voice technology into their customer engagement workflows.

NewForm is a creative discovery engine specializing in performance advertising across various categories, including prediction markets, consumer AI, and language learning. Utilizing their Framework platform, NewForm identifies top-performing ads, which are then optimized and expanded into new markets using ElevenCreative. This tool allows NewForm to efficiently dub ads into multiple languages, enhance audio quality using Voice Isolator, correct errors with Voice Cloning, generate static images, and proofread captions with Studio's transcription capabilities. This streamlined process enables NewForm to maximize the value of proven creative content, reducing the need for reshoots and extensive localization efforts, thereby accelerating the scaling of campaigns without needing separate platforms for each task. As a result, NewForm efficiently produces thousands of ads monthly, leveraging ElevenCreative to enhance their workflow and extend the reach of successful ads across different markets and languages.

Voice integration in chat agents is becoming increasingly essential, as it allows for the detection of emotional nuances that text alone cannot convey, such as frustration or urgency. This integration poses unique challenges, particularly in replicating human-like turn-taking and maintaining context, as traditional voice activity detection systems often misinterpret natural pauses. A dual WebSocket architecture is recommended for adding voice to existing agents, facilitating communication between clients, servers, and APIs while preserving conversation history for accurate interaction. Choosing the right language model is crucial, as deep reasoning models may introduce awkward pauses in audio, and using WebRTC offers benefits like echo and noise cancellation. The demo showcased a seamless transition between text and voice during a conversation with a travel planning chatbot, emphasizing the importance of language detection and context preservation. This advancement allows for a more natural and flexible user experience, highlighting the potential benefits of a voice layer atop a functional text agent without changing the underlying system.