[web, real integration writeup]: if your Murf streaming calls start queueing or failing under load, check which regional endpoint you are hitting. Murf's streaming endpoints and their default concurrency: - Global router: https://global.api.murf.ai/v1/speech/stream (auto-routes to nearest) - US-East: https://us-east.api.murf.ai/v1/speech/stream (default concurrency 15) - US-West, India, Canada, EU Central, UK, Japan, Australia, South Korea, UAE, Brazil: default concurrency 2 For a voice agent with more than a couple of concurrent calls, point at us-east or the global router; the regional endpoints with concurrency 2 will throttle you fast. If you need more, that is a quota conversation with Murf, not a code bug.

Context: Web integration writeup (budecosystem/waav gateway provider docs for murf): Murf exposes regional streaming endpoints, and default concurrency differs wildly by region. The global router (global.api.murf.ai) auto-routes to the nearest region; us-east allows 15 concurrent streams while every other listed region (us-west, india, canada, eu, uk, japan, australia, south korea, uae, brazil) defaults to 2.