Repository navigation
perf(beeline): local call transcription speed on one CPU thread - #376
Merged
Merged
Conversation
Publishes the 2026-11-13 run that fed the beeline 0.1.0 decision to ship local transcription with no GPU and no cloud service: 1644 s of stereo call audio in 222 s of wall time, 7.4x real time, with sherpa-onnx 2.0.0 in WebAssembly on 1 thread and the gigaam-v3 model. The run used the owner's real calls, so the page gives totals only. The script takes the profile and the beeline-cli checkout as arguments. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
The owner wanted to show a potential client a working beeline demo and asked how fast it transcribes calls. The question was whether CPU-only local recognition keeps up with a call center's daily volume. The answer fed the decision to ship local transcription in beeline 0.1.0, with no GPU and no cloud service.
What
performance/beeline/transcription/README.md: why, setup, results, what it showed, not measured.performance/beeline/transcription/run.sh: the script that produced the numbers. The profile and the beeline-cli checkout are arguments.performance/beeline/README.md(new area) and a link fromperformance/README.md. The area keeps no PROCESS.md.Headline: 1644 s of stereo call audio in 222 s of wall time, 7.4× faster than real time, on one CPU thread (sherpa-onnx 2.0.0 in WebAssembly, gigaam-v3). The run used the owner's real calls, so only totals are published.
Testing
pnpm docs:check,pnpm standards:check, rumdl, markdownlint and cspell on the new pages,bash -non the script. The script was not re-run here: it reads a real account.🤖 Generated with Claude Code