Kaldi vs Speechmatics

Kaldi is a free, open-source C++ toolkit for speech recognition aimed at researchers and professionals. Its code and scripts cover methods including linear transforms, MMI, boosted MMI, MCE discriminative training and feature-space discriminative training. The toolkit integrates finite-state transducers through OpenFst and includes a matrix library that wraps standard BLAS and LAPACK routines. Example recipes support building speech recognition systems, and downloadable model archives cover speech recognition, speaker identification, speech activity detection, diarization and language modeling. The models may be used for any purpose. Kaldi is released under Apache License 2.0, which permits commercial use and redistribution. It runs on Linux distributions and Mac OS X, supports Android cross-compilation, and can be compiled to WebAssembly for browser execution. Windows compatibility is not actively maintained, and example scripts do not work there. Users obtain the current development version by cloning the GitHub repository and following its INSTALL file. Documentation is expert-oriented; the project directs help requests to its mailing list and bug reports or feature requests to GitHub issues.

Speechmatics provides APIs for speech-to-text, voice agents, and text-to-speech, processing recorded files or live audio. Speech-to-text supports more than 55 languages, including conversations that switch languages mid-sentence. The company says real-time transcription can return results in less than one second; speaker diarization and custom vocabulary support speaker identification and specialized terms. Listed uses include healthcare, voice agents, live captioning, contact center analytics, legal transcription, and meeting platforms. Deployment options include SaaS, multi-region cloud, on-premises containers and virtual appliances, on-device, and hybrid cloud with on-premises; availability varies by plan. The service is available through API and on Linux, macOS, Windows, web, and self-hosted platforms. The free plan provides $100 in credits, requires no credit card, and allows two concurrent real-time sessions. Usage-based options include text-to-speech at $0.01 per once and transcription plans priced per hour. Speechmatics states it is ISO/IEC 27001:2022 accredited, GDPR compliant, HIPAA compliant, and SOC 2 Type II certified. It says data is encrypted at rest with AES 256 and in transit using TLS 1.2 or higher.