Research Staff - Voice AI Foundations
About the job
- A strong mathematical foundation in statistical learning theory, particularly in areas relevant to self-supervised and multimodal learning.
- Deep expertise in foundation model architectures and an understanding of scaling training across multiple modalities.
- The ability to bridge theory and practice by deriving novel mathematical formulations and implementing them efficiently.
- Demonstrated ability to build data pipelines that process and curate massive datasets while maintaining quality and diversity.
- A track record of designing controlled experiments that isolate the impact of architectural innovations and validate theoretical insights.
- Experience optimizing models for real-world deployment, including knowledge of hardware constraints and efficiency techniques.
- A history of open-source contributions or research publications that advanced the state of the art in speech or language AI.
- Develop next-generation neural audio codecs for extreme low-bit-rate compression and high-fidelity reconstruction across a world-scale general-audio corpus.
- Develop steerable generative models that synthesize diverse human speech from codec latent representations, including casual conversation, emotional expression, multi-speaker scenarios, environmental noise, and overlapping speech.
- Develop embedding systems that factorize codec latent space into interpretable dimensions of speaker, content, style, environment, and channel effects, enabling precise control and amplification of seed datasets through latent recombination.
- Use latent recombination to generate synthetic audio data at scale and train multimodal speech-to-speech systems to understand people across demographics, states, and environments and produce empathic, human-like responses for conversational or task-oriented objectives.
- Design model architectures, training schemes, and inference algorithms adapted to hardware at the bare metal for cost-efficient training on billion-hour datasets and real-time inference at scale.
Company Overview
Deepgram is the leading platform underpinning the emerging trillion-dollar Voice AI economy, providing real-time APIs for speech-to-text (STT), text-to-speech (TTS), and building production-grade voice agents at scale. More than 200,000 developers and 1,300+ organizations build voice offerings that are ‘Powered by Deepgram’, including Twilio, Cloudflare, Sierra, Decagon, Vapi, Daily, Cresta, Granola, and Jack in the Box. Deepgram’s voice-native foundation models are accessed through cloud APIs or as self-hosted and on-premises software, with unmatched accuracy, low latency, and cost efficiency. Backed by a recent Series C led by leading global investors and strategic partners, Deepgram has processed over 50,000 years of audio and transcribed more than 1 trillion words. There is no organization in the world that understands voice better than Deepgram.
Company Operating Rhythm
At Deepgram, we expect an AI-first mindset—AI use and comfort aren’t optional, they’re core to how we operate, innovate, and measure performance.
Every team member who works at Deepgram is expected to actively use and experiment with advanced AI tools, and even build your own into your everyday work. We measure how effectively AI is applied to deliver results, and consistent, creative use of the latest AI capabilities is key to success here. Candidates should be comfortable adopting new models and modes quickly, integrating AI into their workflows, and continuously pushing the boundaries of what these technologies can do.
Additionally, we move at the pace of AI. Change is rapid, and you can expect your day-to-day work to evolve just as quickly. This may not be the right role if you’re not excited to experiment, adapt, think on your feet, and learn constantly, or if you’re seeking something highly prescriptive with a traditional 9-to-5.
The Opportunity
Voice is the most natural modality for human interaction with machines. However, current sequence modeling paradigms based on jointly scaling model and data cannot deliver voice AI capable of universal human interaction. The challenges are rooted in fundamental data problems posed by audio: real-world audio data is scarce and enormously diverse, spanning a vast space of voices, speaking styles, and acoustic conditions. Even if billions of hours of audio were accessible, its inherent high dimensionality creates computational and storage costs that make training and deployment prohibitively expensive at world scale. We believe that entirely new paradigms for audio AI are needed to overcome these challenges and make voice interaction accessible to everyone.
The Role
As a Member of the Research Staff, you will pioneer the development of Latent Space Models (LSMs), a new approach that aims to solve the fundamental data, scale, and cost challenges associated with building robust, contextualized voice AI. Your research will focus on solving one or more of the following problems:
Build next-generation neural audio codecs that achieve extreme, low bit-rate compression and high fidelity reconstruction across a world-scale corpus of general audio.
Pioneer steerable generative models that can synthesize the full diversity of human speech from the codec latent representation, from casual conversation to highly emotional expression to complex multi-speaker scenarios with environmental noise and overlapping speech.
Develop embedding systems that cleanly factorize the codec latent space into interpretable dimensions of speaker, content, style, environment, and channel effects -- enabling precise control over each aspect and the ability to massively amplify an existing seed dataset through “latent recombination”.
Leverage latent recombination to generate synthetic audio data at previously impossible scales, unlocking joint model and data scaling paradigms for audio. Endeavor to train multimodal speech-to-speech systems that can 1) understand any human irrespective of their demographics, state, or environment and 2) produce empathic, human-like responses that achieve conversational or task-oriented objectives.
Design model architectures, training schemes, and inference algorithms that are adapted for hardware at the bare metal enabling cost efficient training on billion-hour datasets and powering real-time inference for hundreds of millions of concurrent conversations.
The Challenge
We are seeking researchers who:
See "unsolved" problems as opportunities to pioneer entirely new approaches
Can identify the one critical experiment that will validate or kill an idea in days, not months
Have the vision to scale successful proofs-of-concept 100x
Are obsessed with using AI to automate and amplify your own impact
If you find yourself energized rather than daunted by these expectations—if you're already thinking about five ideas to try while reading this—you might be the researcher we need. This role demands obsession with the problems, creativity in approach, and relentless drive toward elegant, scalable solutions. The technical challenges are immense, but the potential impact is transformative.
It's Important to Us That You Have
Strong mathematical foundation in statistical learning theory, particularly in areas relevant to self-supervised and multimodal learning
Deep expertise in foundation model architectures, with an understanding of how to scale training across multiple modalities
Proven ability to bridge theory and practice—someone who can both derive novel mathematical formulations and implement them efficiently
Demonstrated ability to build data pipelines that can process and curate massive datasets while maintaining quality and diversity
Track record of designing controlled experiments that isolate the impact of architectural innovations and validate theoretical insights
Experience optimizing models for real-world deployment, including knowledge of hardware constraints and efficiency techniques
History of open-source contributions or research publications that have advanced the state of the art in speech/language AI
Ideal Researcher Profile
An ideal researcher, as evidenced by the recurring themes across these foundational papers, excels in five key areas: (1) Statistical & Mathematical Foundations, (2) Algorithmic Innovation & Implementation, (3) Data-Driven & Scalable Systems, (4) Hardware & Systems Understanding, and (5) Rigorous Experimental Design. Below is a synthesis of how each paper highlights these qualities, with references illustrating why they matter for building robust, impactful deep learning models.
1. Statistical & Mathematical Foundations
Mastery of Core Concepts
Many papers, like Scaling Laws for Neural Language Models and Neural Discrete Representation Learning (VQ-VAE), reflect the importance of power-law analyses, derivation of novel losses, or adaptation of fundamental equations (e.g., in VQ-VAE's commitment loss or rectified flows in Scaling Rectified Flow Transformers). Such mathematical grounding clarifies why models converge or suffer collapse.
Combining Existing Theories in Novel Ways
Papers such as Moshi (combining text modeling, audio codecs, and hierarchical generative modeling) and Finite Scalar Quantization (FSQ's adaptation of classic scalar quantization to replace vector-quantized representations) show how reusing but reimagining known techniques can yield breakthroughs. Many references (e.g., the structured state-space duality in Transformers are SSMs) underscore how unifying previously separate research lines can reveal powerful algorithmic or theoretical insights.
Logical Reasoning and Assumption Testing
Across all papers—particularly in the problem statements of Whisper or Rectified Flow Transformers—the authors present assumptions (e.g., "scaling data leads to zero-shot robustness" or "straight-line noise injection improves sample efficiency") and systematically verify them with thorough empirical results. An ideal researcher similarly grounds new ideas in well-formed, testable hypotheses.
2. Algorithmic Innovation & Implementation
Creative Solutions to Known Bottlenecks
Each paper puts forth a unique algorithmic contribution—Rectified Flow Transformers redefines standard diffusion paths, FSQ proposes simpler scalar quantizations contrasted with VQ, phi-3 mini relies on curated data and blocksparse attention, and Mamba-2 merges SSM speed with attention concepts.
Turning Theory into Practice
Whether it's the direct preference optimization (DPO) for alignment in phi-3 or the residual vector quantization in SoundStream, these works show that bridging design insights with implementable prototypes is essential.
Clear Impact Through Prototypes & Open-Source
Many references (Whisper, neural discrete representation learning, Mamba-2) highlight releasing code or pretrained models, enabling the broader community to replicate and build upon new methods. This premise of collaboration fosters faster progress.
3. Data-Driven & Scalable Systems
Emphasis on Large-Scale Data and Efficient Pipelines
Papers such as Robust Speech Recognition via Large-Scale Weak Supervision (Whisper) and BASE TTS demonstrate that collecting and processing hundreds of thousands of hours of real-world audio can unlock new capabilities in zero-shot or low-resource domains. Meanwhile, phi-3 Technical Report shows that filtering and curating data at scale (e.g., "data optimal regime") can yield high performance even in smaller models.
Strategic Use of Data for Staged Training
A recurring strategy is to vary sources of data or the order of tasks. Whisper trains on multilingual tasks, BASE TTS uses subsets/stages for pretraining on speech tokens, and phi-3 deploys multiple training phases (web data, then synthetic data). This systematic approach to data underscores how an ideal researcher designs training curricula and data filtering protocols for maximum performance.
4. Hardware & Systems Understanding
Efficient Implementations at Scale
Many works illustrate how researchers tune architectures for modern accelerators: the In-Datacenter TPU paper exemplifies domain-specific hardware design for dense matrix multiplications, while phi-3 leverages blocksparse attention and custom Triton kernels to run advanced LLMs on resource-limited devices.
Real-Time & On-Device Constraints
SoundStream shows how to compress audio in real time on a smartphone CPU, demonstrating that knowledge of hardware constraints (latency, limited memory) drives design choices. Similarly, Moshi's low-latency streaming TTS and phi-3-mini's phone-based inference highlight that an ideal researcher must adapt algorithms to resource limits while maintaining robustness.
Architectural & Optimization Details
Papers like Mamba-2 in Transformers are SSMs and the In-Datacenter TPU work show how exploiting specialized matrix decomposition, custom memory hierarchies, or quantization approaches can lead to breakthroughs in speed or energy efficiency.
5. Rigorous Experimental Design
Controlled Comparisons & Ablations
Nearly all papers—Whisper, FSQ, Mamba-2, BASE TTS—use systematic ablations to isolate the impact of individual components (e.g., ablation on vector-quantization vs. scalar quantization in FSQ, or size of codebooks in VQ-VAEs). This approach reveals which design decisions truly matter.
Multifold Evaluation Metrics
From MUSHRA listening tests (SoundStream, BASE TTS) to FID in image synthesis (Scaling Rectified Flow Transformers, FSQ) to perplexity or zero-shot generalization in language (phi-3, Scaling Laws for Neural Language Models), the works demonstrate the value of comprehensive, carefully chosen metrics.
Stress Tests & Edge Cases
Whisper's out-of-distribution speech benchmarks, SoundStream's evaluation on speech + music, or Mamba-2's performance on multi-query associative recall demonstrate the importance of specialized challenge sets. Researchers who craft or adopt rigorous benchmarks and "red-team" their models (as in phi-3 safety alignment) are better prepared to address real-world complexities.
Summary
Overall, an ideal researcher in deep learning consistently demonstrates:
A solid grounding in theoretical and statistical principles
A talent for proposing and validating new algorithmic solutions
The capacity to orchestrate data pipelines that scale and reflect real-world diversity
Awareness of hardware constraints and system-level trade-offs for efficiency
Thorough and transparent experimental practices
These qualities surface across research on speech (Whisper, BASE TTS), language modeling (Scaling Laws, phi-3), specialized hardware (TPU, Transformers are SSMs), and new representation methods (VQ-VAE, FSQ, SoundStream). By balancing these attributes—rigorous math, innovative algorithms, large-scale data engineering, hardware-savvy optimizations, and reproducible experimentation—researchers can produce impactful, trustworthy advancements in foundational deep learning.
Foundational Papers
This job description was generated through analysis of the following papers:
Robust Speech Recognition via Large-Scale Weak Supervision (arXiv:2212.04356)
Moshi: a speech-text foundation model for real-time dialogue (arXiv:2410.00037)
Scaling Rectified Flow Transformers for High-Resolution Image Synthesis (arXiv:2403.03206)
Scaling Laws for Neural Language Models (arXiv:2001.08361)
BASE TTS: Lessons from building a billion-parameter Text-to-Speech model on 100K hours of data (arXiv:2402.08093)
In-Datacenter Performance Analysis of a Tensor Processing Unit (arXiv:1704.04760)
Neural Discrete Representation Learning (arXiv:1711.00937)
SoundStream: An End-to-End Neural Audio Codec (arXiv:2107.03312)
Finite Scalar Quantization: VQ-VAE Made Simple (arXiv:2309.15505)
Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone (arXiv:2404.14219)
Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality (arXiv:2405.21060)
Notice: We're aware of individuals impersonating Deepgram recruiters. All legitimate Deepgram recruiting communication comes from an @deepgram.com email address. If you've received a message claiming to be Deepgram, please forward it to [email protected].
About the company
Deepgram builds speech software for developers and businesses that turn spoken conversations into text, generate voices, and run automated phone agents. Its Nova recognition models process live audio and recordings, with tools to identify speakers and recognize industry terms. Developers send audio or text through its APIs and receive transcripts, spoken responses, or extracted information that their applications use in customer service and other workflows. The company sells usage priced by audio minutes or generated characters, with prepaid credits and enterprise contracts for larger deployments. Deepgram was founded in 2015 and is headquartered in San Francisco, California, United States.
Company Size
51-200
Company Stage
Series C
Total Funding
$233.3M
Headquarters
San Francisco, California
Founded
2015
Get referred to Deepgram
See people who can refer or advise you
Simplify's Take
What believers are saying
- Expansion has fresh financing: Deepgram raised $130 million in January 2026 to develop its voice platform and expand internationally.
- Regional demand is rising: Deepgram reported 96% annual growth in APAC API requests in August 2026.
- US engineers have geographic flexibility: Deepgram lists remote inference and voice-agent engineering roles across the United States.
What critics are saying
- Faster responses consume more model calls: Deepgram says Flux's optional early predictions add 50-70% more language-model calls.
- Customer deployments add security work: Deepgram's self-hosted containers leave tenant authentication and log separation to operators.
- Inference engineers face changing requirements: Deepgram's role posting warns against expecting a highly prescriptive 9-to-5.
What makes Deepgram unique
- Flux combines two jobs: one model transcribes speech and decides when a speaker has finished.
- Customers control where conversations stay: Deepgram's self-hosted deployments keep audio and transcripts inside customer infrastructure.
- Its efficiency work has patent protection: Deepgram's speech recognition patent covers batching and parallel processing.
Help us improve and share your feedback! Did you find this helpful?
Benefits
- Flexible Work Hours
Growth & Insights and Company News
6 month growth
0%1 year growth
0%2 year growth
-4%Genesys enhances Agentic Virtual Agent amid growing enterprise adoption. Faster development, advanced AI reasoning, new voice innovations and deepened enterprise connectivity enable virtual agents to act across business systems and resolve more complex customer requests LAS VEGAS - Sept. 2, 2026 - Today at Xperience, Genesys(R), a global leader in agentic orchestration for customer experience (CX), announced advancements to Genesys Cloud(TM) Agentic Virtual Agent that will help organizations automate more complex customer interactions with greater speed, reliability and natural engagement. New enterprise connectivity and interoperability will also expand the ability of Agentic Virtual Agent to securely work with business systems and other AI agents to drive customer outcomes across the enterprise. Since becoming generally available last March, continual enhancements for Agentic Virtual Agent are making it easier to build, test and optimize autonomous customer experiences alongside the new Scaled Cognition APT-2 large action model as well as new native voice capabilities and integrations with ElevenLabs and Deepgram. Agentic Virtual Agent will also provide deeper Model Context Protocol (MCP) connectivity and Agent2Agent (A2A) interoperability with enterprise platforms. Organizations including Global Payments, MTN South Africa, Utility Warehouse, UOL and Charles Sturt University are choosing Agentic Virtual Agent to advance beyond traditional automation. For some organizations, that impact is translating into multimillion-dollar annualized savings. Customers are also reporting improved satisfaction and Net Promoter Score gains of more than 70 points, with some deployments going live in as little as a few weeks. Increasing the reliability of automated experiences Genesys is expanding the intelligence behind Agentic Virtual Agent through the latest Scaled Cognition APT-2 large action model, designed to improve reasoning accuracy, factual grounding and execution reliability. APT-2 supports richer agent specifications, longer and more complex workflows, faster response times and improved multilingual performance. These enhancements help Agentic Virtual Agent reason through sophisticated customer requests and carry work toward resolution within enterprise-defined guardrails, all while helping reduce experience failures that can erode trust. Creating best-in-class voice experiences New voice capabilities for Agentic Virtual Agent are designed to enable more natural, responsive conversations. Through End-of-Turn Detection, Agentic Virtual Agent will understand when a customer has finished a thought, helping conversations flow with fewer awkward pauses and interruptions. With Contextual Tuning, Agentic Virtual Agent will automatically adapt its terminology, tone and speaking style to reflect an organization's business and brand as well as the context of each interaction. Genesys is also providing greater customer choice through new integrations with ElevenLabs and Deepgram across voice generation and conversational speech recognition. * Through ElevenLabs, organizations will be able to access expressive, natural-sounding voices for engaging virtual agent interactions while maintaining Genesys Cloud governance and enterprise controls. * Deepgram will bring real-time speech recognition built for conversational AI, helping Agentic Virtual Agent understand interruptions, turn-taking and changing intent as people speak. Richer, more timely context gives the agent stronger inputs for reasoning and action, supporting faster responses and more reliable task completion. These new capabilities can make voice interactions with Agentic Virtual Agent feel more fluid and natural, helping organizations deliver engaging customer experiences that increase successful self-service and reduce unnecessary escalations to human agents. Building trusted agentic experiences faster Across the development lifecycle, organizations can more rapidly build and deploy Agentic Virtual Agent experiences. With spec-driven development, teams can describe desired customer experiences and outcomes using existing materials such as process documentation, standard operating procedures and interaction transcripts, which AI helps translate into deployable configurations. Development teams can also use a range of tools such as Anthropic's Claude Code, OpenAI Codex, Cursor and Kiro to create specifications that can be brought into Genesys Cloud for testing and governance, providing flexibility while maintaining enterprise oversight. Genesys is introducing new AI-assisted authoring that analyzes how Agentic Virtual Agents are designed, identifies opportunities to improve their effectiveness and recommends ways to drive better customer outcomes. New AI-powered testing, expanded reporting and detailed auditability will also help organizations validate Agentic Virtual Agents before deployment, understand their actions and continuously improve performance over time. Creating the most connected Agentic Virtual Agent Building on orchestration capabilities from its recent acquisition of Pinkfish, Genesys is expanding Agentic Virtual Agent with more than 500 enterprise integrations and access to over 25,000 MCP-compatible tools, making it easier to connect to the business context, systems and tools behind customer interactions. This will enable Agentic Virtual Agent to securely access CRM, ERP, IT service management, knowledge, collaboration, billing and other business applications to gather context and take action to help resolve customer requests. Progressing multi-agent enterprise Action A2A interoperability will enable Agentic Virtual Agent to securely collaborate with specialized AI agents across platforms such as Salesforce and ServiceNow, delegating tasks and coordinating work across multistep business processes. Combined with MCP-based access to enterprise tools, A2A will connect customer conversations to the systems and AI agents needed to reach resolution. For example, Agentic Virtual Agent will be able to coordinate with specialized AI agents across enterprise platforms to resolve a billing issue, from retrieving account information and evaluating policies to completing backend tasks, while maintaining context throughout the customer interaction. Agentic Virtual Agent will also support customer engagement across digital channels, including WhatsApp, enabling organizations to bring agentic experiences to customers through the channels they choose while A2A and MCP provide interoperability with the enterprise systems and AI agents behind those interactions. Global customer momentum Around the world, organizations are realizing measurable business impact with Agentic Virtual Agent, including: * Riachuelo, one of Brazil's largest fashion retailers, replaced its traditional chatbot with Genesys Agentic Virtual Agent, increasing customer retention from 30% to 84%, improving its CSAT by 60 points within the first year and boosting productivity by more than 400% without adding staff. * Zeppelin Rental GmbH, a leading provider of rental and construction support services, is enhancing its digital customer experience with the Genesys Agentic Virtual Agent. By enabling customers to describe their job site needs in natural language instead of having to search through product catalogs, the solution makes relevant products, services and information easier to find. This creates a simpler, more intuitive self-service experience while helping to reduce pressure on traditional service channels. "Enterprise AI is entering a new phase defined by what it can accomplish for customers," said Mike Szilagyi, senior vice president and head of product at Genesys. "We're further strengthening Genesys Cloud Agentic Virtual Agent with more context, intelligence and capabilities to act across the enterprise, work with other AI agents and carry increasingly complex customer requests through resolution. This is how businesses can turn customer intent into meaningful outcomes at scale." Genesys Cloud Agentic Virtual Agent capabilities announced today, including the Scaled Cognition APT-2 model, new development tools and the Deepgram integration, are available now with ElevenLabs expected to be available in the third quarter of the company's fiscal year (Aug 1-Oct. 31, 2026). Additional capabilities, including native voice enhancements, A2A interoperability and AI-assisted authoring, are expected to become generally available in the fourth quarter of the company's fiscal year (Nov. 1, 2026-Jan. 31-2027). Forward-Looking Statements Statements in this press release that are not historical or current facts are forward-looking statements that involve risks and uncertainties. Unless required by law, Genesys undertakes no obligation to update or revise any forward-looking statements to reflect circumstances or events after the date of this press release. About Genesys Genesys(R) empowers more than 8,000 organizations worldwide to create the best customer and employee experiences. Genesys Cloud(TM) is the Agentic Orchestration Platform that securely connects people, systems, data and AI across the enterprise with built-in governance and control. As a result, organizations can drive customer loyalty, growth and retention while increasing operational efficiency and teamwork across human and AI workforces. To learn more, visit www.genesys.ai. (C) 2026 Genesys. All rights reserved. Genesys, the Genesys logo, Genesys Cloud, Genesys Cloud CX, GCXNow, Experience as a Service, AppFoundry and Genesys Orchestrators are trademarks, service marks and/or registered trademarks of Genesys. All other company names and logos may be registered trademarks or trademarks of their respective companies. Media Contact
India’s Ringg gets backing from Peak XV as it pushes voice AI past the phone call. - August 25, 2026 More than 76% of consumers in India prefer talking to businesses over a phone call, according to a recent study from Truecaller. Since voice is still consumers' preferred way to communicate, that leaves a big opportunity to automate support and outreach calls using voice AI in the country. Voice AI startup Ringg, which already processes 20 million call attempts a month, is betting that volume keeps climbing over the coming months, and it just raised more money on that belief. The company said today it has landed $10 million from Peak XV Partners as an extension of its Series A. It had previously raised $5.5 million in a Series A round earlier this year, bringing the round's total to $15.5 million. Ringg started life as a text-to-speech startup called DesiVocal, but training its own speech models proved expensive, so the founders moved up the stack - building voice AI agents for enterprises instead. Indian fintech Cred became its first customer, and Ringg has since signed Indian startups like Flipkart, Practo, Groww, and PolicyBazaar. "At the start, we were doing high-volume, low-complexity use cases like outbound calling, lead qualification, loan collection, and more. We quickly realized these are not sticky use cases, and so it's always going to be a price game," the startup's co-founder Siddharth Tripathi told TechCrunch. Ringg still serves some of those simpler use cases, but it has set its sights on more complex workflows: appointment booking for healthcare clinics, abandoned-cart recovery for e-commerce sites, and onboarding/KYC ("know your customer") checks for fintech apps. Tripathi said Ringg's voice agent now runs across 1,200 clinics for the healthcare app Practo, helping patients book visits or follow up on next steps post-visit. Voice calls still make up over 70% of Ringg's business, but the startup has started branching into other channels, including chat and WhatsApp. For some clients like Shell, it's also automating browser-based support requests. Why is the DOJ investigating Andreessen Horowitz over board seats? Equity Podcast 0 seconds of 39 minutes, 2 seconds Volume 0% "We are trying to position ourselves as a platform for agents that bring outcomes or get things done rather than voice agents for enterprises," Tripathi said. Most of Ringg's customers are based in India, with a handful in the Middle East and the U.S. But the startup isn't trying to sell directly to U.S. companies; instead, it wants to partner with so-called Global Capability Centers in India - the offshore hubs multinationals increasingly lean on for back-office and support work - to sell automation capacity alongside human support. Tripathi said the company builds its own speech recognition and generation models, and would eventually like to own the full voice stack, including infrastructure and deployment. For now, though, that remains too costly, so the product works as an orchestration layer, routing tasks to different models depending on the use case. Rishen Kapoor, a principal at Peak XV, said that because Ringg started as a research lab building its own models, that technical depth shows up in the complex use cases it's now tackling. "Because of the technical capabilities, they can actually do these hard-won enterprise workflows end to end. They can complete these higher-value tasks like merchant onboarding, like L1 and L2 support, with quality and with consistency," Kapoor told TechCrunch. Voice AI in India is a crowded field. Model makers including Deepgram, ElevenLabs, Cartesia, and local players like Sarvam and Smallest.ai, are all jockeying for pole position. Orchestration-focused startups like Bolna and Blue Machines are chasing the same layer Ringg occupies, while sector-focused players like Gnani and Arrowhead concentrate heavily on finance. That layered stack - model makers, orchestrators, and application-layer players all trying to lock in enterprise workflows - is itself the story. The money and the defensibility increasingly sit with whoever owns the customer relationship and the outcome. Ringg currently has 40 employees, with more than 15 hired in the last three months. The startup is hiring for forward-deployed engineer roles that combine technical chops with product management skills, along with researchers focused on bringing down the cost of running its models.
Deepgram speech-to-text is now free for all Vagle AI users. Every Vagle AI account now includes Deepgram streaming speech-to-text at no additional cost. No separate Deepgram API key or billing is required. Direct answer. Starting today, Deepgram real-time speech-to-text is included at no additional cost for every Vagle AI user. You do not need your own Deepgram API key, and there is no separate per-minute Deepgram charge on your account. This applies to both browser web calls and phone calls made through the platform. Vagle covers the Deepgram transcription cost as part of its calling infrastructure, backed by a Deepgram investment in the platform. What changed. Previously, real-time transcription providers were billed as a separate cost component on each connected call minute. Users needed to understand which STT provider was selected and what it cost. Now, Deepgram streaming transcription is bundled into the platform. Users selecting Deepgram as their transcriber see no extra line item - the cost is absorbed by Vagle. What you get. * Real-time streaming speech-to-text on every call * Low-latency interim and final transcripts * Support for English and multiple languages * Automatic voice activity detection (VAD) * Works on browser web calls and telephony (Twilio, Plivo, VoBiz) * No separate Deepgram API key needed * No per-minute Deepgram billing on your account How to use it. When you create or edit an assistant in Vagle, select Deepgram as the transcriber. That is all - there is no API key to paste and no billing toggle to enable. If you already use Deepgram as your transcriber, nothing changes in your workflow. The cost simply disappears from your usage breakdown. Why this is possible. Deepgram has invested in Vagle AI, which includes infrastructure credits that allow the platform to offer their transcription at no additional user cost. This is a direct result of that backing. As long as the investment credits cover usage, Deepgram STT remains free for users. If this changes in the future, users will be notified in advance. Does this affect other providers. No. If you prefer a different transcriber (Groq Whisper, Speechmatics, or another supported option), those remain available with their standard pricing. Deepgram is the one that is now included at no extra cost. Limits and fair use. There is no hard per-minute cap announced today. Standard platform fair-use policies apply. Extremely high-volume usage that exceeds what the investment credits cover may be subject to future adjustments, with advance notice. Faq. Do I need a Deepgram API key to use this? Is Deepgram STT really free on Vagle? Does this work for phone calls too? Can I still use other transcribers? Will this remain free forever? Official provider resources. Related resources. Tell Vagle your use case, call volume, tools, and follow-up needs. Vagle will help you choose a practical setup.
Voice AI infrastructure leader Deepgram to establish APAC headquarters in Singapore following strategic investment from EDBI. Deepgram to deepen support across APAC, where it serves customers in more than 20 markets; names former Salesforce leader Sriram Ved to lead the region. Aug 18, 2026 Prev Next 1 of 43,493 Deepgram, the real-time AI infrastructure company underpinning the Voice AI economy, announced plans to establish its Asia-Pacific headquarters in Singapore, deepening its commitment to a region where the company is already seeing strong demand for production-grade Voice AI. Deepgram has appointed Sriram Ved as Vice President and General Manager for APAC who will lead the company's regional strategy and oversee its customer, partner, and go-to-market expansion. Deepgram saw a 96% year-over-year increase in API requests across APAC and serves customers across more than 20 markets in the region, including Australia, China, Hong Kong, India, Indonesia, Japan, Singapore, South Korea, Thailand, and Vietnam. From Singapore, the company plans to bring regional leadership, customer support, partnerships, and commercial operations closer to customers across the region, build on its existing speech-to-text expertise and support across Asian languages, and expand its low-latency inference footprint. The move follows a strategic investment from EDBI, an arm of SG Growth Capital, the investment platform of EDB and Enterprise Singapore, as part of an extension to Deepgram's US$130 million Series C funding. EDBI will also support Deepgram as it establishes a presence in Singapore and builds commercial relationships and regional operations across APAC. Companies are moving Voice AI into high-volume, customer-facing operations spanning contact centers, financial services, healthcare, telecommunications, and commerce in APAC. As deployments scale, enterprises need infrastructure that can perform reliably across languages, accents, and noisy environments while meeting strict requirements for latency, cost, privacy, and data control. Deepgram plans to deepen its regional investment in three core areas: * Customer and partner support: Expand its regional leadership, customer support, partnerships, and commercial operations, bringing these teams and capabilities closer to customers across APAC. * Asian-language capabilities: Build on Deepgram's existing speech-to-text support for all four of Singapore's official languages: English, Mandarin Chinese, Malay, and Tamil, while deepening its expertise in Asian languages, accents, and speech patterns, including multilingual interactions, code-switching, and specialized terminology. Deepgram's broader speech-to-text portfolio currently supports Japanese, Hindi, Thai, Cantonese, and several Indic languages and regional variants. * Low-latency, flexible deployment: Work toward a broader low-latency inference footprint across APAC while continuing to offer cloud, virtual private cloud, self-hosted, and on-premises deployments that help enterprises meet regional requirements around privacy, data control, and data residency. "APAC is one of the most promising markets for Voice AI right now," said Scott Stephenson, CEO and co-founder of Deepgram. "You have huge demand, incredible linguistic diversity, and companies moving quickly from experimenting with AI to putting it into production. Singapore puts us closer to the customers and partners building that future, and gives us a strong base to help make Voice AI work across the complexity of the region." "Voice AI is already live in production across APAC, from contact centers and financial services to healthcare and telecommunications," said Sriram Ved, Vice President and General Manager for APAC at Deepgram. "We're also seeing a new generation of AI companies building for the region and turning to Deepgram for the voice infrastructure underneath their products. They need technology that works where real conversations happen: across noise, accents, languages, and massive volumes, without sacrificing speed or reliability. That's what Deepgram was built to do." EDBI invests in global companies that drive the development of innovative solutions, create good jobs, and contribute to Singapore's long-term economic resilience. Its investment in Deepgram reflects a shared ambition to build advanced AI capabilities from Singapore while supporting the company's growth across APAC. "Voice AI is becoming critical infrastructure for enterprises across Asia. The companies developing these technologies need to be close to the markets, languages, and deployment realities they serve. Deepgram's decision to anchor its APAC headquarters here reflects a long-term commitment to building from Singapore. At EDBI, we partner with companies like Deepgram to establish and scale meaningful capabilities here, and we look forward to its continued contribution to Singapore's AI ecosystem," said Yeung Chia Li, Senior Partner at EDBI. Globally, Deepgram powers more than 200,000 developers and 1,400 organizations. Its platform has processed more than 50,000 years of audio and transcribed more than one trillion words.
Voice AI infrastructure leader Deepgram to establish APAC headquarters in Singapore following strategic investment from EDBI. * admin@cb26 * August 18, 2026 * 0 Comments * 41 Views Business Wire India Deepgram, the real-time AI infrastructure company underpinning the Voice AI economy, announced plans to establish its Asia-Pacific headquarters in Singapore, deepening its commitment to a region where the company is already seeing strong demand for production-grade Voice AI. Deepgram has appointed Sriram Ved as Vice President and General Manager for APAC who will lead the company's regional strategy and oversee its customer, partner, and go-to-market expansion. Deepgram saw a 96% year-over-year increase in API requests across APAC and today serves customers across more than 20 markets in the region, including Australia, China, Hong Kong, India, Indonesia, Japan, Singapore, South Korea, Thailand, and Vietnam. From Singapore, the company plans to bring regional leadership, customer support, partnerships, and commercial operations closer to customers across the region, build on its existing speech-to-text expertise and support across Asian languages, and expand its low-latency inference footprint. The move follows a strategic investment from EDBI, an arm of SG Growth Capital, the investment platform of EDB and Enterprise Singapore, as part of an extension to Deepgram's US$130 million Series C funding. EDBI will also support Deepgram as it establishes a presence in Singapore and builds commercial relationships and regional operations across APAC. Companies are moving Voice AI into high-volume, customer-facing operations spanning contact centers, financial services, healthcare, telecommunications, and commerce in APAC. As deployments scale, enterprises need infrastructure that can perform reliably across languages, accents, and noisy environments while meeting strict requirements for latency, cost, privacy, and data control. Deepgram plans to deepen its regional investment in three core areas: * Customer and partner support: Expand its regional leadership, customer support, partnerships, and commercial operations, bringing these teams and capabilities closer to customers across APAC. * Asian-language capabilities: Build on Deepgram's existing speech-to-text support for all four of Singapore's official languages: English, Mandarin Chinese, Malay, and Tamil, while deepening its expertise in Asian languages, accents, and speech patterns, including multilingual interactions, code-switching, and specialized terminology. Deepgram's broader speech-to-text portfolio currently supports Japanese, Hindi, Thai, Cantonese, and several Indic languages and regional variants. * Low-latency, flexible deployment: Work toward a broader low-latency inference footprint across APAC while continuing to offer cloud, virtual private cloud, self-hosted, and on-premises deployments that help enterprises meet regional requirements around privacy, data control, and data residency. "APAC is one of the most promising markets for Voice AI right now," said Scott Stephenson, CEO and co-founder of Deepgram. "You have huge demand, incredible linguistic diversity, and companies moving quickly from experimenting with AI to putting it into production. Singapore puts us closer to the customers and partners building that future, and gives us a strong base to help make Voice AI work across the complexity of the region." "Voice AI is already live in production across APAC, from contact centers and financial services to healthcare and telecommunications," said Sriram Ved, Vice President and General Manager for APAC at Deepgram. "We're also seeing a new generation of AI companies building for the region and turning to Deepgram for the voice infrastructure underneath their products. They need technology that works where real conversations happen: across noise, accents, languages, and massive volumes, without sacrificing speed or reliability. That's what Deepgram was built to do." EDBI invests in global companies that drive the development of innovative solutions, create good jobs, and contribute to Singapore's long-term economic resilience. Its investment in Deepgram reflects a shared ambition to build advanced AI capabilities from Singapore while supporting the company's growth across APAC. "Voice AI is becoming critical infrastructure for enterprises across Asia. The companies developing these technologies need to be close to the markets, languages, and deployment realities they serve. Deepgram's decision to anchor its APAC headquarters here reflects a long-term commitment to building from Singapore. At EDBI, we partner with companies like Deepgram to establish and scale meaningful capabilities here, and we look forward to its continued contribution to Singapore's AI ecosystem," said Yeung Chia Li, Senior Partner at EDBI. Globally, Deepgram powers more than 200,000 developers and 1,400 organizations. Its platform has processed more than 50,000 years of audio and transcribed more than one trillion words. Deepgram is moving forward with plans to open its Singapore office as part of its long-term investment in the region. Learn more at deepgram.com. About Deepgram Deepgram is the real-time AI infrastructure company underpinning the Voice AI economy. Today, more than 200,000 developers and 1,400 organizations are Powered by Deepgram. Its Voice AI platform offers speech-to-text (STT), text-to-speech (TTS), and full speech-to-speech (STS) capabilities, all powered by an enterprise-grade runtime. Deepgram's voice-native foundation models, accessed through cloud APIs or as self-hosted/on-premises APIs, deliver unmatched accuracy, low latency, and competitive pricing. Customers include technology ISVs building voice products or platforms, co-sell partners working with large enterprises, and enterprises solving internal use cases. Having processed over 50,000 years of audio and transcribed over 1 trillion words, there is no organization in the world that understands voice better than Deepgram. To learn more, please visit www.deepgram.com, read its developer docs, or follow @DeepgramAI on X and LinkedIn.