[LauraGPT]

Final Publication Acceptance

GitHub Release and PyPI 1.4.24 are both published. Both distributions were downloaded from official PyPI and match the accepted GitHub assets by SHA-256 and size:

• funasr-1.4.24-py3-none-any.whl: 1015161 bytes; SHA-256 c2678ebb77c7dfba1d85a79749ad30b441159b87d17eff901e0c6ad4ebcc06c7.
• funasr-1.4.24.tar.gz: 1081873 bytes; SHA-256 522dcb493be75aa57d8183e6197b4431c5e02fade689eda90cfd91f76e5edb57.

The official sdist download resumed after a transport timeout using an exact HTTP 206 range; the original partial and failure receipts were retained. The wheel was not downloaded again and PyPI was not re-uploaded.

FunASR 1.4.24

Source: 20f26b7c4905d010768ec91206149df2a90104b6 / tag v1.4.24.

Fix

Deliver #3852: ignore empty non-linguistic symbols instead of allowing the character/phoneme tokenizer scan to stall. Deliver #3853: correct variance normalization for padded utterances. Deliver #3854: preserve valid CIF output and gradients when a batch mixes utterances with zero and nonzero fires, including dynamic ONNX batches.

Compatibility: checkpoints trained with both norm_means=True and norm_vars=True require recognition-quality reevaluation after the corrected normalization. norm_vars=False remains the default. Local checkpoint round trips are not a pretrained recognition-quality evaluation.

This does not claim pretrained ASR quality, Windows/NPU acceptance or a full repository-suite result.

The separately released funasr-onnx==0.4.9 and funasr-torch==0.1.6 have their own tags, packages and verification. This root-package release does not republish them. PR #3848 is not included. This release does not claim every EOS path is fixed.

Verification Scope

• New candidate release/README checks: 45 cases, 0 skips. Related docs/package/release checks: 104 cases, 2 explicitly recorded skips; no failures or errors.
• Actual installed candidate and final wheels: 37 passing CPU/ONNX cases each covering tokenizer symbols, variance normalization, local checkpoint round trips, repository-audio inference and CIF dynamic batches. The 27 CUDA parametrizations are intentionally skipped locally and separately covered by bound prior hardware evidence. All 439 package files and 13 entrypoints match the accepted source.
• Reused 24 tokenizer/CTC cases, 34 FP32 CIF cases and four normalizer repository-audio cases including six CUDA inference calls on L20Z, after rehashing 33,870 sealed evidence files and 18 helpers and binding the exact source to the installed wheel. The 106-case normalizer acceptance retains 102 unchanged cases and four repaired audio checks. Earlier harness/TF32-fixture failures remain preserved; they are not silently discarded or counted as passes. This release performs no new GPU execution and makes no pretrained-quality or full-Paraformer-export claim.
• The local candidate build and candidate/final package checks, tests and installations use private HOME/cache directories and separate network namespaces with verified network-unreachable probes. The final-tag build runs in GitHub Actions under the existing release workflow, not the local network namespace. No newly downloaded pretrained weights are used. The separately reported fix(auto): keep a result for inputs whose VAD speech decodes to no text #3848 test-download incident is not an authorization or validation input for this release.
• Release-head CI was accepted from actual logs and recorded artifact inventories. The roadmap job checked out the pure head; the other head workflows used GitHub's temporary PR merge commit. Both checkout trees were verified against the exact candidate. Merged-main and final-tag source checkouts were separately verified. The main API-doc workflow also checked out only the retained Chinese API page from gh-pages into previous-pages; this separate documentation input was recorded, not treated as inference source. The newly downloaded Python wheel/sdist passed byte/metadata/archive/source checks, strict Twine, offline installation, actual import and CLI checks.
• The 10 unchanged runtime archives reuse complete cache bytes freshly hashed against new asset metadata and the existing runtime source release. They were not rebuilt or re-executed. SHA256SUMS covers all 12 payloads. This is not a new runtime-hardware claim.
• No full repository-suite, production-environment upgrade, Nano benchmark or adoption/growth claim is made.

GitHub Release and official PyPI downloads are both accepted; the plan below is retained history.

Historical Pre-Release Plan

Summary

Release the already merged fixes #3852, #3853 and #3854 as FunASR 1.4.24:

• Ignore empty non-linguistic symbols to prevent tokenizer scan loops.
• Correct padded utterance variance normalization.
• Handle mixed zero-fire CIF batches, including the tested dynamic ONNX path.

Checkpoints trained with both norm_means=True and norm_vars=True require recognition-quality reevaluation because corrected normalized features differ. norm_vars=False remains the default. This does not claim pretrained checkpoint quality.

Only the version, release contract, four README release summaries and two repository-role version links change here. Inference source is unchanged from accepted main b5d0fd98926682c9f804d51f7787e6f488435bfd.

Completed Candidate Verification

• Release-contract RED: 8 expected failures and 4 passes before the metadata update; GREEN release/README checks: 45 passed.
• Related documentation/package/release checks: 102 passed, 2 explicitly recorded local skips. Hosted browser/product checks remain required.
• Acceptance-helper safety contracts: 25 passed.
• Candidate wheel/sdist build and strict Twine checks passed. Offline installation, actual imports, CLI, 439 package-file bytes and 13 entrypoints verified.
• Installed wheel: 31 tokenizer/normalizer CPU cases passed; 27 CUDA parametrizations intentionally skipped. Six CIF CPU/ONNX methods passed with no skips. These controlled local checkpoint/audio fixtures are not pretrained quality tests.
• Prior sealed L20Z evidence was freshly rehashed (33,870 evidence files and 18 helpers), then exact inference-source hashes were bound to the installed wheel: 24 tokenizer/CTC cases, 34 FP32 CIF cases, and four normalizer repository-audio cases including six CUDA inference calls. The 106-case normalizer acceptance and its earlier negative results remain preserved. No new GPU run is claimed.
• Local builds, tests and installations used private HOME/cache directories and kernel network isolation. No newly downloaded pretrained weights were used. The separate fix(auto): keep a result for inputs whose VAD speech decodes to no text #3848 download incident is excluded.

Remaining Release Gates

Actual release-head CI, normal merge, merged-main CI and final-tag CI must each be accepted independently. Final GitHub wheel/sdist must be downloaded and verified, installed-wheel behavior rechecked, and all 12 payloads covered by a read-back SHA256SUMS manifest (13 assets total). The 10 unchanged runtime archives may reuse complete caches only after fresh full-hash, asset-metadata and archive-integrity validation; they are not new builds or hardware runs. PyPI upload is last, followed by actual official downloads matching GitHub bytes.

The standalone ONNX/Torch packages are not republished. No full repository-suite, Windows/NPU, production upgrade, Nano benchmark or adoption-growth claim is made.