Expired50% confidenceBenchmarkExact time
Cloud-based STT services typically take 200–500ms, while locally deployed models such as Whisper can be compressed to under 100ms with GPU acceleration.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/14/2026
First Seen
Valid until: 10/12/2026(expired)
Sources
状态机语音Agent实战：Pipecat与Vapi的延迟与准确性权衡
redditr/aiagents7/12/2026
Related Claims
UnverifiedLocal deployment latency consists of three segments: faster-whisper STT on GPU approximately 100–300 ms, first token from a 7B-parameter LLM approximately 200–500 ms, and TTS synthesis approximately 100–300 ms.76% similarUnverified业界领先的STT实时转录延迟约在200-400毫秒，采用Whisper或DeepSpeech等模型72% similarVerifiedThe serverless GPU cold start process for large models—allocating an instance, pulling the image, and loading weights—can take 30 seconds or even several minutes.71% similarUnverifiedCloud TTS services typically have a time-to-first-byte of 200–500 milliseconds, while locally deployed lightweight TTS models (such as Piper and Kokoro-82M) can reduce latency to under 50 milliseconds, though audio quality is generally slightly lower.68% similarVerifiedCommand execution latency on a local area network can be as low as under 5 milliseconds, far superior to the 50–300 millisecond round-trip latency of cloud-based solutions.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/503986API
curl https://kongchang.com/api/v1/knowledge/claims/503986MCP
get_claim(id=503986)