Murf AI — Complete Company Research

Identity

What Murf solves

• Turns scripts into professional AI voiceovers without conventional recording logistics.
• Supports multilingual voice generation and localization.
• Provides API infrastructure for embedded / automated speech generation.
• Provides streaming TTS for conversational AI and voice agents.
• Provides enterprise governance and security controls for business deployments.

Sneha Roy — Co-Founder & COO

[VERIFIED] Murf's legal/official and investor materials identify Sneha Roy as a co-founder; public company profiles identify her as COO.

Elevation Capital published a podcast episode with Roy covering the Murf idea, early days, GTM, PLG, integrations, synthetic media expansion and team building.

Education: IIT Kharagpur is documented by Elevation and TechCrunch as the founders' shared alumni background.

Divyanshu Pandey — Co-Founder / Chief Business Officer

[VERIFIED] Current public company profiles identify Divyanshu Pandey as a co-founder and business leader. His public LinkedIn post confirms his participation in the 2022 Series A announcement alongside Edkie and Roy.

Education: University of Delaware is listed by Crunchbase's current profile. [THIRD-PARTY REPORTED]

Current management evidence

Murf's March 2026 Impressum names Ankur Edkie, Sneha Roy and Divyanshu Pandey under Management. This is the strongest current source located for the leadership trio.

Additional executive titles beyond this trio are not treated as current unless supported by a current primary source.

Falcon 2 — current streaming focus

[VERIFIED] Murf's API documentation describes Falcon 2 as an ultra-fast, scalable speech synthesis model designed for real-time conversational AI. The documentation states ~100ms time-to-first-audio and support for thousands of concurrent sessions; Murf's product page separately reports a 95ms live 30-day median TTFA across production calls.

• Streaming TTS for voice agents and live interactions.
• Current API documentation: 150+ voices, 35 languages, 20+ speaking styles across the Murf API ecosystem.
• Regional endpoints include US, India, Canada, South Korea, UAE, Japan, Australia, EU Central, UK and South America.
• Enterprise custom voice cloning is documented on the Falcon 2 product page and in the September 2026 product update.

Gen2 — studio-quality / non-streaming focus

[VERIFIED] Gen2 is described as Murf's realistic and customizable speech model for studio-quality synthesis. Current documentation says the legacy Gen2 streaming model is deprecated; Gen2 remains available through the non-streaming Synthesize Speech endpoint.

• 70,000+ hours of diverse, ethically sourced audio claimed by Murf.
• Native 44.1 kHz output described in documentation.
• Strong control over rhythm, emotion, pronunciation and multilingual identity consistency.
• Independent blind-test language evaluation is described by Murf; the 99.38% API benchmark is still a company-published evaluation, not a customer-success metric.

Voice library

Current Murf pages cite 200+ voices across 35+ languages on Studio/enterprise-facing pages, while the API documentation cites 150+ voices across 35 languages. These figures are retained separately rather than merged because the product surfaces define their libraries differently.

Voice cloning

Murf currently describes consent-based voice cloning. Falcon 2's product page says enterprise users can clone a custom brand voice from 10 seconds of spoken audio with explicit consent. A September 8, 2026 product update says enterprise customers can create API voice clones from a short recording of up to 30 seconds, with multilingual output across Falcon 2 locales. These are product-surface-specific claims and should not be treated as one universal recording requirement.

Voice Changer

The API exposes a voice-changer conversion endpoint. Murf describes it as transforming recordings into AI voices while preserving delivery/emotional characteristics; current API documentation supports file-based conversion.

AI Dubbing

Murf's current API page advertises dubbing in 25+ languages with linguistic review. The product ecosystem also includes translation and localization workflows. Exact language counts vary by product surface and should be quoted with date/context.

Documented total funding

Combining the two publicly documented rounds. No later primary funding announcement was found in this review through September 24, 2026.

Current valuation

Third-party databases may display estimates; this report does not treat those estimates as a verified current valuation.

Documented motions

• Freemium / free trial entry point.
• Self-serve Studio plans.
• Enterprise sales and custom contracts.
• API/developer acquisition.
• Integrations with Canva, PowerPoint, Google Slides, Articulate and Adobe tools.
• Customer case studies and named customer-leader campaigns.
• Startup incubator credits.
• Voice-actor partnerships and ethical-AI positioning.

Analyst interpretation

[INFERENCE] The product architecture suggests a dual GTM: PLG/self-serve for creators and smaller teams, combined with enterprise sales around governance, localization, API infrastructure and high-volume voice-agent deployments.

[INFERENCE] The migration from Studio-centric messaging to “voice infrastructure” and Falcon 2 suggests an expansion from content creation into embedded speech and conversational AI infrastructure.

What is publicly disclosed

• Gen2 documentation says the model was trained on 70,000+ hours of diverse and ethically sourced audio.
• Murf says its voices come from real people who choose to share their talent.
• Murf says it uses transparent contracts and compensates professional voice actors through its voice-actor model.
• Murf publishes an Ethical AI page and copyright policy.

What is not publicly disclosed in sufficient detail

• A complete itemized training-data inventory.
• All datasets and licenses underlying every model generation.
• Full model architecture details.
• A complete public red-team / abuse-testing dataset for voice cloning.

Training-data details beyond disclosed summaries: Not publicly disclosed.

Product / technical limitations

• Gen2 streaming was deprecated in 2026, requiring migration to Falcon 2 for streaming use cases.
• Different product surfaces publish different voice-count and pricing units; users must verify exact plan/model scope.
• AI voice quality and pronunciation are benchmark-dependent; Murf's 99.38% figure is a company evaluation, not proof of universal speech accuracy.

Market / governance risks

• Voice cloning raises consent, impersonation and rights-management concerns across the industry.
• Speech models are increasingly commoditized by cloud and specialist providers.
• Enterprise deployments may require human review for regulated, safety-critical or brand-sensitive content.
• Public customer ROI data is selective and primarily company/customer-reported.

Strengths

• Broad voiceover Studio workflow plus API infrastructure.
• Current streaming model strategy through Falcon 2.
• Multilingual voice, translation and dubbing capabilities.
• Enterprise security/compliance positioning.
• Named customer evidence in L&D, aviation, agency and industrial settings.

Weaknesses

• Public financial disclosure is limited.
• Customer metrics are selective and often first-party.
• Model/product naming has changed as streaming technology evolved.
• Some public marketing metrics are difficult to independently validate.

Opportunities

• Voice agents and conversational AI.
• Enterprise localization and dubbing.
• Embedded speech APIs.
• Regulated enterprise workflows requiring governance.

[INFERENCE] These are market opportunities derived from current product direction.

Threats

• Fast-moving competition from specialist TTS, voice-agent and cloud platforms.
• Model commoditization and pricing pressure.
• Voice-cloning regulation and rights disputes.
• Enterprise procurement requirements around security and data residency.

[INFERENCE] Threat framing is analytical, not a claim about an internal Murf problem.

Important facts

• Founded in October 2020 with three founders, ten voices and three clients.
• Documented funding: $1.5M seed + $10M Series A.
• Current legal entity: Murf Inc.
• Current headquarters: San Francisco.
• Falcon 2 is the current streaming TTS model.
• Gen2 streaming is deprecated; non-streaming remains.
• Murf markets both content-production and conversational AI infrastructure.

Key lessons for AI companies

• Enterprise expansion is supported by solving a recurring workflow bottleneck, not just generating better audio.
• Localization can turn TTS into a global operations product.
• Integrations reduce workflow friction and support adoption inside existing content tools.
• Moving from a point product to infrastructure/API can broaden the addressable market.
• Ethical voice sourcing and enterprise governance become strategic product attributes as voice cloning scales.

Strong fit profile [INFERENCE]

• Organizations producing recurring voice content at scale.
• Global L&D teams with multilingual requirements.
• Agencies and production teams needing frequent revisions.
• Developers building real-time voice experiences.
• Enterprises that need security, governance and regional API controls.
• Teams where recording, studio scheduling and voice talent coordination are recurring bottlenecks.

Potentially lower need [INFERENCE]

• Projects requiring a single bespoke human performance where production volume is very low.
• Organizations with no recurring voice-generation or localization workflow.
• Highly specialized productions where human actors, union agreements or live direction are mandatory.
• Use cases where AI-generated voice is not acceptable to the audience or regulator.