# Meta enters real-time speech recognition market with Muse Voice Transcribe New model supporting 20+ speakers priced at just $0.18 per hour targets enterprise customers. - **Source:** [VentureBeat](https://venturebeat.com) - **Published:** 2026-09-02 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/6bb05f94-0e96-4316-99fb-220037e2239b ## Meta's new speech recognition model Meta announced Muse Voice Transcribe, a new real-time speech recognition model capable of identifying 20+ speakers. The model features long-form audio processing and multilingual support. Priced at just $0.18 per hour, it presents a cost-effective solution for businesses. This move positions Meta as a significant competitor in the speech recognition market. ## The importance of diarization technology Diarization technology identifies who is speaking, not just what is said. Unlike traditional speech recognition, it determines speaker identity, crucial for meeting systems, customer service analytics, and compliance processes. Meta's Muse model integrates diarization directly into its architecture for improved accuracy in complex scenarios. ## Competitive pricing strategy At $0.18 per hour, Meta's pricing undercuts many competitors like Amazon Transcribe and Speechmatics. This aggressive pricing particularly targets SMBs and startups, potentially expanding market share. Meta offers zero-data-retention processing at the same price point, addressing privacy-conscious enterprises. ## Technical innovations Muse Voice Transcribe features adaptive delay mechanisms, waiting longer on ambiguous speech while committing quickly to clear audio. Trained on 70+ languages with extensive validation for 25, it handles multilingual code-switching and keyword biasing. The model processes audio in 80-millisecond chunks at 12.5 chunks per second. ## Competitive landscape While Speechmatics supports up to 100 speakers and Amazon Transcribe handles 30, Meta's 20+ capability positions it in the mid-high segment. Meta's differentiation combines capacity with low-latency transcription and competitive pricing. AssemblyAI's 10-speaker limit shows the varying capabilities across providers in this evolving market. ## Enterprise use cases Ideal for meeting systems, live assistants, call analytics, and ambient AI, Muse particularly benefits multi-participant virtual meetings and customer service calls. Its diarization accuracy could significantly reduce post-meeting processing time. GDPR-compliant zero-data-retention processing appeals to regulated industries. ## Architectural details Muse uses an autoregressive multimodal architecture processing audio in 80ms chunks. At each step, it decides whether to consume more audio or emit text. Meta employs reinforcement learning combining word-error-rate and delay rewards to train this behavior, optimizing both accuracy and responsiveness. ## API capabilities Meta's Model API documentation positions diarization as a first-class operation mode alongside push-to-talk and endpointing. The API manages speaker labels session-wise and provides turn-level timestamps. Parity pricing between standard and zero-retention processing demonstrates Meta's focus on privacy-conscious development. ## Market impact and future Meta's entry may drive down prices and raise quality standards in speech recognition. Making this technology accessible to SMBs could accelerate automation adoption. The competitive pricing pressures rivals to innovate, potentially benefiting the entire industry through improved offerings and more accessible solutions. ## Sources covering this story - https://venturebeat.com/technology/meta-prices-muse-voice-transcribe-at-0-18-an-hour-with-real-time-diarization-for-20-speakers-a-steal-for-enterprises — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._