Unverified50% confidenceBenchmarkExact time
UP host's actual test: FasterWhisper's Small model took 70 seconds to transcribe 10 minutes of audio, and the Base model took 28 seconds; the full 35 minutes 02 seconds took 253.3 seconds to transcribe, about 8.3x real time, outputting 1,209 subtitles.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/29/2026
First Seen
Valid until: 1/27/2027
Sources
Related Entities
Related Claims
UnverifiedEnd-to-end models have reduced average latency in speech translation from the 2–3 seconds of traditional pipelines to approximately 300 milliseconds.79% similarUnverifiedThe local Whisper model transcribes audio at approximately 1/4 to 1/10 of real-time speed on consumer-grade GPUs, meaning processing 35,000 hours of content could take several months.78% similarUnverifiedOn a 16G GPU, a 60-second audio clip takes only 101 seconds with 512-frame chunking, 231 seconds with 1024-frame chunking, and 285 seconds without chunking.78% similarUnverifiedOído's firmware runs on a single core at 1.97 times the speed of audio duration, meaning a 4-second instruction can only be transcribed approximately 8.7 seconds after it is finished being spoken.78% similarUnverifiedIn parallel mode, the first audio arrived only 2.3 to 2.5 seconds after the request (rather than 1.0 to 1.2 seconds), and in the three runs all stutters were concentrated in the first 13 seconds (while the LLM was still writing or had just finished), with no stutters afterward.77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1110109API
curl https://kongchang.com/api/v1/knowledge/claims/1110109MCP
get_claim(id=1110109)