Coqui

Coqui began as a continuation of the open speech tooling lineage that originated around Mozilla Common Voice and the TTS/STT research community; the company was founded in 2020 and positions itself as an open-source-first vendor for speech technology. Its core value proposition is delivering production-quality TTS and STT while keeping model code, checkpoints, and training pipelines accessible so teams can run inference locally or on private cloud. Coqui balances an open-source library approach (Coqui TTS and Coqui STT projects) with a commercial hosted API and managed services for teams that prefer not to operate inference infrastructure.

Feature-wise, Coqui provides several concrete capabilities. Its TTS stack supports neural architectures such as GlowTTS/FastSpeech-style pipelines and VITS-style end-to-end models with HiFi-GAN vocoder support for high-fidelity audio; users can synthesize multi-lingual speech and export 16/24-bit WAVs. Coqui also offers voice cloning/fine-tuning workflows that can produce usable mimic voices from a few minutes of curated audio (voice adaptation workflows commonly report usable output from ~1–5 minutes of clear speech). On the STT side, Coqui STT offers offline transcription models trained on Common Voice data and supports streaming recognition for low-latency use cases. For production, Coqui supplies a hosted REST API with streaming endpoints, SDKs (Python/Node), and Docker images for local deployment.

On pricing, Coqui maintains the open-source code and model checkpoints freely available for local use; that core level is effectively free to run (infrastructure costs still apply). For hosted convenience, Coqui offers a pay-as-you-go API and managed plans; hosted access and volume tiers are quoted on a usage basis and enterprise contracts are custom. There is typically a free trial or limited free tier for API testing, while sustained production usage moves customers to billed API usage or a negotiated enterprise plan. Specific hosted API quotas, overage rates, and enterprise SLAs are provided during signup or by contacting sales for custom pricing.

Who uses Coqui? Real-world workflows include a localization engineer training accented TTS voices for a multilingual app, and a podcast producer cloning short guest reads to speed episode production. Specifically, a backend engineer uses Coqui to deploy an on-premise TTS microservice for 100k monthly minutes of IVR audio, and an accessibility specialist uses STT models to generate captions meeting GDPR constraints. In head-to-head terms, Coqui is most often compared to hosted-first vendors like ElevenLabs; the practical difference is Coqui’s open-source artifacts and on-prem deployment options versus ElevenLabs’ fully hosted voice library approach.