modelscope/FunASR
1
0
Fork 0
mirror of https://github.com/modelscope/FunASR.git synced 2026-10-12 02:19:18 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(websocket): finalize file input on end control (#3440)

Browse Source 
Signed-off-by: LauraGPT <[email protected]>
Co-authored-by: LauraGPT <[email protected]>
This commit is contained in:
zhifu gao
2026-07-29 20:24:01 +08:00
committed by GitHub
co-authored by LauraGPT
parent 6570924ed0
commit 30caaf8e18
4 changed files with 938 additions and 32 deletions
Download Patch File Download Diff File Expand all files Collapse all files
runtime/python/websocket/README.md
+4 -1
View File
@@ -68,9 +68,12 @@ python funasr_wss_client.py \
--audio_in [if set, loadding from wav.scp, else recording from mircrophone] \
--output_dir [if set, write the results to output_dir] \
--mode [`online` for streaming asr, `offline` for non-streaming, `2pass` for unifying streaming and non-streaming asr] \
--thread_num [thread_num for send data]
--thread_num [thread_num for send data] \
--result_timeout [seconds to wait for the final server acknowledgement]
```
When `--audio_in` is set, the client sends `{"is_speaking": false, "is_end": true}` after the last audio frame. The server flushes pending online and offline inference before replying with `{"is_end": true, "is_final": true}`. If inference fails, the acknowledgement contains `{"is_end": true, "is_final": false, "error": "..."}`. The client waits up to `--result_timeout` seconds for this acknowledgement; the default is 300 seconds.
#### Usage examples
##### ASR offline client
Recording from mircrophone