* The accuracy is excellent, even on noisy audio or with multiple speakers. Many of the transcripts required minimal editing.
* Speaker diarisation works reliably — being able to split out who said what is a big plus in multi-person recordings.
* Ease of integration is a standout: the API is well documented, the onboarding is smooth, and I got up and running quickly.
* The pricing model is fair and transparent — you pay for usage rather than being locked into a subscription.
* Advanced features like Word Boost / keyword prompting, PII redaction, and language auto-detection give useful flexibility for real-world use cases. Review collected by and hosted on G2.com.
* The latency/response times can vary under load, which makes it less predictable for real-time needs.
* Customisation is somewhat limited: fine-tuning for domain-specific vocabulary or acoustic quirks isn’t as deep as one might hope.
* The API returns many fields in the response; for simpler workflows, that extra metadata can add overhead.
* The 10-hour audio length limit (for some endpoints) feels restrictive for very long recordings.
* In certain regions (e.g. Europe), some features are either missing or still in development. Review collected by and hosted on G2.com.
