nvidia
/

stt_uz_fastconformer_hybrid_large_pc

Automatic Speech Recognition

hf-asr-leaderboard

Model card Files Files and versions

rima-shahbazyan commited on Oct 31, 2024

Commit

6c9d45b

·

verified ·

1 Parent(s): cf471db

Update README.md

Files changed (1) hide show

README.md +5 -11

README.md CHANGED Viewed

@@ -144,17 +144,11 @@ The performance of Automatic Speech Recognition models is measuring using Word E
 The following tables summarizes the performance of the model with the Transducer decoder. Performances of the ASR models are reported in terms of Word Error Rate (WER%) with greedy decoding.
-|     **Tokenizer**     | **Vocabulary Size** | **MCV 17.0 TEST** | **KSD TEST** | **KSC2 TEST Read** | **KSC2 TEST Spontaneous** |
-|:---------------------:|:-------------------:|:-----------------:|:------------:|:------------------:|:-------------------------:|
-| SentencePiece Unigram |         1024        |       15.48       |      7.08    |        4.43        |            15.25          |
-|                           |                        **WER(%)**                            |
-|---------------------------|:------:|:------------:|:--------------:|:-------------------:|
-|                           |**WER** |**WER wo CAP**|**WER wo PUNCT**| **WER wo PUNCT CAP**|
-|---------------------------|:------:|:------------:|:--------------:|:-------------------:|
-| FLEURS DEV (used as test) |  17.52 |        16.20 |          12.20 |               10.73 |
-| MCV TEST                  |  16.46 |        15.89 |           7.78 |                7.18 |
 ## Limitations
 The model is non-streaming and outputs the speech as a string without capitalization and punctuation. Since this model was trained on publicly available speech datasets, the performance of this model might degrade for speech which includes technical terms, or vernacular that the model has not been trained on.

 The following tables summarizes the performance of the model with the Transducer decoder. Performances of the ASR models are reported in terms of Word Error Rate (WER%) with greedy decoding.
+|---------------------------|:--------:|:------------:|:--------------:|:-------------------:|
+|                           |**WER(%)**|**WER wo CAP**|**WER wo PUNCT**| **WER wo PUNCT CAP**|
+|---------------------------|:--------:|:------------:|:--------------:|:-------------------:|
+| FLEURS DEV (used as test) |    17.52 |        16.20 |          12.20 |               10.73 |
+| MCV TEST                  |    16.46 |        15.89 |           7.78 |                7.18 |
 ## Limitations
 The model is non-streaming and outputs the speech as a string without capitalization and punctuation. Since this model was trained on publicly available speech datasets, the performance of this model might degrade for speech which includes technical terms, or vernacular that the model has not been trained on.