sanchit-gandhi HF staff commited on
Commit
de6ec26
1 Parent(s): c7a80b8

Update constants.py

Browse files
Files changed (1) hide show
  1. constants.py +1 -1
constants.py CHANGED
@@ -60,7 +60,7 @@ Example: If it takes an ASR system 10 seconds to transcribe 10 seconds of speech
60
  If it takes 20 seconds to transcribe the same 10 seconds of speech, the RTF is 2.
61
  ```
62
 
63
- For the benchmark, we report RTF averaged over a 10 minute audio sample with 5 warm up batches followed 3 graded batches.
64
 
65
  ## How to reproduce our results
66
 
 
60
  If it takes 20 seconds to transcribe the same 10 seconds of speech, the RTF is 2.
61
  ```
62
 
63
+ For the benchmark, we report RTF averaged over a 10 minute audio sample that is chunked into 30 second segments, mimicking the [chunked long-form transcription strategy](https://huggingface.co/blog/asr-chunking) performed in Transfomrmers. We measure RTF on an A100 80GB GPU (Driver Version: 535.54.03, CUDA Version: 12.2), performing 5 warm-up runs and 3 graded runs, over which the RTF is averaged to get the final result.
64
 
65
  ## How to reproduce our results
66