theadamsabra nielsr HF Staff commited on
Commit
3306270
·
verified ·
1 Parent(s): 12a5996

Update GitHub repository link (#1)

Browse files

- Update GitHub repository link (0c91a9d2ed90b726b551ad9e6e7b92db174e87dc)


Co-authored-by: Niels Rogge <[email protected]>

Files changed (1) hide show
  1. README.md +7 -6
README.md CHANGED
@@ -1,14 +1,15 @@
1
  ---
2
- license: other
3
  language:
4
- - zh
5
  library_name: transformers
 
6
  pipeline_tag: automatic-speech-recognition
7
- arxiv: https://arxiv.org/abs/2509.02523
8
  ---
 
9
  # Moonshine
10
 
11
- [[Paper]](https://arxiv.org/abs/2509.02523) [[Installation]](https://github.com/usefulsensors/moonshine/blob/main/README.md)
12
 
13
  This is the model card for running the automatic speech recognition (ASR) models (Moonshine models) trained and released by Moonshine AI (f.k.a Useful Sensors.)
14
 
@@ -92,7 +93,7 @@ Our evaluations show that, the models exhibit greater accuracy on standard datas
92
 
93
  However, like any machine learning model, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself.
94
 
95
- In addition, the sequence-to-sequence architecture of the model makes it prone to generating repetitive texts, which can be mitigated to some degree by beam search and temperature scheduling but not perfectly. It is likely that this behavior and hallucinations may be worse for short audio segments, or segments where parts of words are cut off at the beginning or the end of the segment.
96
 
97
  ## Broader Implications
98
 
@@ -113,4 +114,4 @@ If you benefit from our work, please cite us:
113
  primaryClass={cs.CL},
114
  url={https://arxiv.org/abs/2509.02523},
115
  }
116
- ```
 
1
  ---
 
2
  language:
3
+ - zh
4
  library_name: transformers
5
+ license: other
6
  pipeline_tag: automatic-speech-recognition
7
+ arxiv: https://arxiv.org/abs/2509.02523
8
  ---
9
+
10
  # Moonshine
11
 
12
+ [[Paper]](https://arxiv.org/abs/2509.02523) [[Code]](https://github.com/moonshine-ai/moonshine)
13
 
14
  This is the model card for running the automatic speech recognition (ASR) models (Moonshine models) trained and released by Moonshine AI (f.k.a Useful Sensors.)
15
 
 
93
 
94
  However, like any machine learning model, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself.
95
 
96
+ In addition, the sequence-to-sequence architecture of the model makes it prone to generating repetitive texts, which can be mitigated to some degree by beam search and temperature scheduling but not perfectly. It is likely that this behavior and hallucinations may be worse for short audio segments, or segments where parts of words are cut off at the beginning or at the end of the segment.
97
 
98
  ## Broader Implications
99
 
 
114
  primaryClass={cs.CL},
115
  url={https://arxiv.org/abs/2509.02523},
116
  }
117
+ ```