Sanas raised a $65M Series B in January 2025 at a reported $555M post-money valuation. Insight Partners led the round, with participation from General Catalyst, Quadrille Capital, and Human Capital.

Before the Series B, Sanas raised a $32M Series A in 2023 led by General Catalyst, following an earlier seed round. Sanas has raised over $100M in total funding.

Sanas provides real-time speech AI that modifies how a contact-center agent sounds on a live call without changing the content of their speech. Its core product, Accent Translation, converts an agent's spoken English accent in real time so that, for example, a caller in the United States hears a locally familiar accent while the agent's natural voice characteristics remain intact. The system processes speech with latency low enough that neither party notices a delay during a normal conversation.

Contact centers integrate Sanas into their existing telephony or communications stacks, with the system processing speech in the audio path. Agents do not need to install desktop software or change their workflows.

Sanas also offers Speech Enhancement, which removes background noise, echo, and other audio artifacts; Language Translation, which allows agents and callers who speak different languages to converse in real time, with each hearing the other in their own language; and Speech Intelligence, which analyzes live calls for sentiment, compliance flags, and conversation insights.

Through its developer platform and SDK, Sanas also allows communications platforms, telecom carriers, and software developers to embed its capabilities directly into their products rather than purchase packaged contact-center applications. Sanas therefore sells both applications and infrastructure.

Sanas sells enterprise subscriptions on a per-user, per-year basis through a B2B go-to-market model. Its public AWS Marketplace listing prices 100 Accent Translation users at $72,000 for a 12-month contract, while Speech Enhancement costs $18,000 for 100 users. Large deployments use negotiated pricing, and management has reported average contract values of approximately $1M.

Gross margins are roughly 90%. The primary cost drivers are GPU inference compute and R&D for training and maintaining low-latency speech models across accents and languages.

Expansion revenue comes from product adoption and seat growth. Customers that start with Accent Translation can add Speech Enhancement, Language Translation, or Speech Intelligence, contributing to roughly 160% net revenue retention. Successful pilots can also expand across an enterprise's agent population, increasing seat count within the same contract. With roughly 90% gross margins and approximately $1M average contract values, this expansion can increase revenue from existing customers without a proportional increase in sales costs.

Sanas operates at the intersection of real-time speech infrastructure and contact-center applications, where horizontal AI platforms, specialist speech companies, and full-stack CCaaS vendors offer overlapping products.

Deepgram, Speechmatics, and OpenAI offer speech recognition and generation capabilities that overlap with parts of Sanas's stack. Deepgram has expanded from speech-to-text into text-to-speech and voice-agent infrastructure, offering developers a unified voice platform. OpenAI's real-time voice API can handle translation and conversation, while distribution through ChatGPT and the API gives it broader reach than startups can build through brand alone.

ElevenLabs and Cartesia focus on speech generation and voice cloning, adjacent categories to accent conversion. ElevenLabs reached an estimated $90M ARR in late 2024, indicating the revenue scale available to voice AI companies. These horizontal providers do not offer a production-grade, real-time accent conversion product tuned for contact-center telephony, giving Sanas a narrower product focus.

Cresta offers real-time voice translation as part of a contact-center AI suite that includes coaching, QA, and conversation intelligence. Genesys, NICE, and other CCaaS incumbents can bundle multilingual voice features into platform renewals, using existing procurement relationships to displace point solutions.

A large CCaaS vendor could build or acquire accent and noise-reduction capabilities, then offer them as included features rather than separately priced modules. Sanas's defense rests on the technical difficulty of real-time accent conversion at telephony-grade latency, along with 200-plus enterprise deployments that provide reference customers and integration depth.

DeepL is expanding real-time speech translation through integrations with meeting and communications platforms, targeting the same multilingual use case from a different angle. HeyGen and Synthesia focus on video dubbing and localization, competing for adjacent enterprise budgets tied to multilingual customer experiences.

Sanas operates in real time on live calls rather than on pre-recorded or asynchronous content. Live phone calls have stricter latency and speaker-identity requirements than batch dubbing, limiting how quickly localization-first companies can enter Sanas's core use case.

Sanas started as a single-product accent tool for BPO call centers and is expanding along three axes: product breadth, customer verticals, and infrastructure distribution.

The addition of Speech Enhancement, Language Translation, and Speech Intelligence extends Sanas from a point solution to a multi-product speech platform. Each product creates a cross-sell opportunity within the existing customer base, with net revenue retention at 160%.

Language Translation addresses enterprises serving customers across language boundaries, beyond BPOs with offshore agents. The combined suite may also increase switching costs by requiring customers to replace multiple integrated capabilities rather than a single feature.

Sanas gained initial traction among BPOs and outsourced customer support providers, but its customer base now includes healthcare, financial services, travel, retail, and telecommunications. Each vertical has distinct compliance, integration, and workflow requirements, allowing Sanas to build domain-specific deployment experience.

The global contact-center market employs tens of millions of agents, compared with Sanas's current 200-plus enterprise customers. Expansion into regions where agents commonly speak English as a second language while serving English-speaking customers, including Southeast Asia, Latin America, and Eastern Europe, would increase the addressable seat count without requiring new product development.

Sanas's SDK and developer platform allow communications platforms, telecom carriers, and UCaaS vendors to embed its speech AI directly into their products. This expands the business model from selling seats to individual enterprises to providing speech capabilities across third-party platforms.

A single carrier integration can provide access to millions of end users without requiring a direct sales relationship with each enterprise. This distribution model also extends Sanas beyond contact centers into telemedicine, remote education, and enterprise collaboration, where accent intelligibility and real-time translation are applicable.

Platform bundling: Large cloud and CCaaS vendors such as Genesys, Microsoft, and Amazon can bundle speech enhancement and translation into existing platform contracts at marginal cost, potentially commoditizing Sanas's core capabilities before it reaches the scale needed to compete on distribution.

Concentration in contact centers: Most of Sanas's revenue comes from contact-center deployments, and a structural industry shift, whether AI-driven call deflection reduces live agent volumes or onshoring trends narrow the accent gap, could compress the addressable market for its core product.

Model commoditization: As open-source and commercial speech models from OpenAI, Deepgram, and others improve real-time accent handling and multilingual translation, Sanas's technical differentiation in latency and speaker-identity preservation may narrow, shifting competition toward price and distribution, where larger platforms have structural advantages.