Miso Labs

Name: Miso Labs
Brand: Miso Labs

AI Native

The most emotive foundation models for voice. Backed by Y Combinator (P26).

Est 2025|Raised $500K|1 Ref Customers

LinkedIn 638|Web traffic 11K|FTEs 3|HQ San Francisco, California, US

Published Sep 27, 2026

01Common complaints

- Company sizes or stages where this product is over-capacity or insufficient - For small applications that need basic narration, commodity cloud provider TTS may be more cost-effective. - For large enterprises with strict procurement gates, the absence of public compliance documentation may impede procurement. - Industries or use cases where competitors are clearly more suitable - Multilingual contact centers or global consumer products: vendors with established multilingual offerings are likely safer choices. - Dubbing/localization pipelines: platforms with mature dubbing workflows and tooling are likely a better fit. - Technical requirements that this product cannot meet - Non-English production support using the open model (MisoTTS is English-only). - Budget constraints that make alternatives more appropriate - Organizations that cannot afford GPU infrastructure for self-hosting and prefer to avoid per-minute API billing may select cheaper character-based commodity TTS alternatives. - Specific workflows or integrations where the product falls short - If an organization requires comprehensive, out-of-the-box voice agent tooling (telephony integrations, evaluation suites, conversation analytics), other platforms with more end-to-end infrastructure may be preferable. - Red flags in the buying process - If a 110 ms target is a strict requirement, procurement should require: - Clear definition of the measured metric (TTFB vs first audio vs full utterance) - Benchmark methodology and measurement conditions - Tail latency measurements under expected load - Hardware and network assumptions used for the benchmarks - Current limitations acknowledged by the vendor / project documentation - English-only support is explicitly stated in the GitHub README. - Latency marketing can be misinterpreted; the project documentation clarifies the 110 ms reference pertains to hosted API TTFB on high-end hardware, not local inference. - Features on the roadmap (not yet available) - Vendor communications indicated API access was forthcoming as of June 3, 2026. Pricing and API plans appear published; buyers should verify current availability and any waitlist constraints. - Scale limitations - No hard public concurrency limits were found. - For self-hosting, scaling 8B models typically requires GPU memory planning, throughput benchmarking, and multi-GPU serving strategies. - Security or compliance gaps - No public SOC 2 or ISO evidence was found in the reviewed sources. - API limitations or integration constraints - The API was not fully assessed; procurement should request full API documentation and test streaming behavior, interruption handling, and long-context consistency.

02Why they buy

- Competitive advantages corroborated by technical documentation (not just marketing claims) - Direct third-party review data is limited. GitHub documentation includes explicit latency caveats and system requirement guidance that provide practical context for engineering evaluation. - Unique features and capabilities relative to closed vendors - Availability of open weights and a local deployment option distinguishes Miso from some fully closed vendors. - Differentiation from competitors based on available information - There were insufficient clearly attributable third-party reviews to claim differentiation based on customer review data. - Areas where competitors have advantages - Brand maturity, documentation depth, multilingual support, and broader review footprints are stronger for several competitors (e.g., ElevenLabs).

03Field reports

Review platform availability (important caveat)

- Direct, attributable end-customer reviews for Miso Labs' voice model were difficult to locate on major B2B review platforms. - Gartner Peer Insights: no specific Miso Labs product page with reviews was found in accessible results. - G2 / TrustRadius: searches for the vendor name can map to unrelated products due to name collisions; no clearly attributable listing or reviews were found during the review session. - Conclusion: In August 2026, public review coverage appears sparse for this vendor relative to more established competitors.

Community / social proof (limited but present)

Hacker News: a submission about the Miso TTS model appears in listings (early June 2026); the thread could not be opened during retrieval, so commenter content could not be quoted directly.
GitHub: the MisoTTS repository shows open-source activity and repository indicators and includes technical caveats that provide context beyond marketing statements.

Examples of quote-level feedback from project documentation

- The project's GitHub documentation contains technical notes useful for procurement and engineering evaluation, including: - "Latency note: Public discussions of a published low-latency figure refer to the hosted production API's time-to-first-byte (TTFB) on high-end accelerator hardware, not to the unoptimized local inference path in this repository. Expect materially slower startup and generation latency on consumer or workstation hardware." - "Language support: Miso TTS 8B currently supports English only."

Inferred reasons for adoption

Teams prioritizing:
- Low TTS time-to-first-byte for streaming agents.
- Availability of open weights and options for local deployment.
- Expressive speech output (emotion/intonation) rather than flat narration.

Inferred reasons for non-adoption or churn

Organizations that require:
- Mature compliance documentation (SOC 2, ISO) prior to procurement.
- Proven uptime SLAs and multi-region hosted deployments.
- Multilingual voice coverage at launch.
- A longer track record of public customer references and reviews.

Public case studies

No detailed, independently verifiable customer case studies were found in the accessible sources reviewed.

Vendor-published demos and samples

Vendor launch materials include demos and vendor-selected samples that illustrate emotive speech capabilities; these materials are vendor-controlled and do not constitute independent validation.

Recommended due diligence requests for procurement

Request a proof-of-concept that includes the customer's own:
- Accent distribution
- Domain vocabulary
- Telephony codecs (e.g., 8 kHz PSTN) versus wideband
- Interruptibility / barge-in behavior
- Latency distribution under load
- Safety controls and watermarking behavior

04Best fit

- Company sizes: specific employee counts and revenue ranges - Likely best fit: - Early-stage and small-to-mid-market teams building voice experiences where user-experience differentiation is prioritized over formal compliance documentation up front. - Larger enterprises seeking on-prem deployments if vendor-provided compliance and security controls can be validated through the sales process. - Industries and verticals where adoption is most relevant - Potentially relevant: - Customer service automation (voice agents) - Scheduling/booking call automation - Consumer applications where expressive voice UX is a differentiator - Potentially less suitable (without enterprise contract and documentation): - Heavily regulated sectors that require documented controls prior to vendor selection - Roles and personas - Decision makers - Head of AI/ML, VP Engineering, CTO - Contact center technology owners for voice agent deployments - Security/IT leadership for on-prem deployments - End users - Applied ML engineers - Voice/agent platform engineers - Product engineers integrating telephony and streaming - Pain points the product addresses - Robotic or low-emotion TTS output that reduces engagement. - Latency-related pauses in voice agents that degrade conversational flow. - Data sovereignty and on-premises processing requirements for voice data. - Qualification criteria: what indicates a strong fit - Recommended benchmarking with realistic workloads should validate: - Latency distribution (typical and tail latency) - Quality with target accents and domain-specific scripts - Operational reliability under concurrency - The organization should have accelerated compute (e.g., GPUs) and MLOps competence if self-hosting is planned. - Deal disqualifiers: what indicates a poor fit - Requirements for immediate multilingual production across many languages. - Requirement for public compliance attestations prior to procurement. - Need for a long track record of public customer reviews and case studies.

05The pitch

What the product does (core value proposition in objective terms)

- Miso Labs provides emotive foundation models for voice designed to produce expressive text-to-speech (TTS) output for real-time voice agents. - The company states that Miso optimizes for: - Low response latency under specific benchmarked conditions. - "One-shot" voice cloning from a brief audio sample. - Open-weight / local deployment options to enable keeping voice processing on-premises. - Miso’s open-weight model is published as MisoTTS (a multi-billion-parameter open-weight model) with inference code available on a public code repository and model weights hosted on a popular model hub.

Target market and market positioning

- Primary buyers/users appear to include: - Developer teams building voice agents (customer support automation, scheduling/booking, outbound calls, concierge agents). - Enterprise and security-focused teams that require on-premises deployment or data sovereignty for voice processing. - Differentiation is presented primarily on latency and expressiveness (emotion/intonation) rather than broad language coverage or a full platform suite of features.

Get a comprehensive analysis of Miso Labs including market position, competitive landscape, adoption trends, and peer benchmarks.

Download full report →