Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -26,20 +26,26 @@ ui.theme = "peach"
|
|
26 |
ui.article = """<h2>Pre-trained model Information</h2>
|
27 |
<h3>Automatic Speech Recognition</h3>
|
28 |
<p style='text-align: justify'>The model used for the ASR part of this space is from
|
29 |
-
https://huggingface.co/facebook/wav2vec2-base-960h which is pretrained and fine-tuned on
|
|
|
30 |
Librispeech</b> on 16kHz sampled speech audio. This model has a <b>word error rate (WER)</b> of <b>8.6 percent on
|
31 |
noisy speech</b> and <b>5.2 percent on clean speech</b> on the standard LibriSpeech benchmark. More information can be
|
32 |
-
found on its website at https://ai.facebook.com/blog/wav2vec-20-learning-the-structure-of-speech-from-raw-audio
|
33 |
-
|
|
|
34 |
<h3>Text Translator</h3>
|
35 |
-
<p style='text-align: justify'>The English to Spanish text translator pre-trained model is from
|
36 |
-
mt-en-es
|
37 |
-
|
|
|
|
|
38 |
cases for many languages. </p>
|
39 |
<h3>Text to Speech</h3>
|
40 |
-
<p style='text-align: justify'> The TTS model used is from
|
|
|
41 |
This model uses the <b>Fairseq(-py)</b> sequence modeling toolkit for speech synthesis, in this case, specifically TTS
|
42 |
-
for Spanish. More information can be seen on their git at
|
|
|
43 |
"""
|
44 |
|
45 |
|
|
|
26 |
ui.article = """<h2>Pre-trained model Information</h2>
|
27 |
<h3>Automatic Speech Recognition</h3>
|
28 |
<p style='text-align: justify'>The model used for the ASR part of this space is from
|
29 |
+
[facebook/wav2vec2-base-960h](https://huggingface.co/facebook/wav2vec2-base-960h) which is pretrained and fine-tuned on
|
30 |
+
<b>960 hours of
|
31 |
Librispeech</b> on 16kHz sampled speech audio. This model has a <b>word error rate (WER)</b> of <b>8.6 percent on
|
32 |
noisy speech</b> and <b>5.2 percent on clean speech</b> on the standard LibriSpeech benchmark. More information can be
|
33 |
+
found on its website at [wav2vec](https://ai.facebook.com/blog/wav2vec-20-learning-the-structure-of-speech-from-raw-audio)
|
34 |
+
and
|
35 |
+
original model is under [pytorch/fairseq](https://github.com/pytorch/fairseq/tree/main/examples/wav2vec).</p>
|
36 |
<h3>Text Translator</h3>
|
37 |
+
<p style='text-align: justify'>The English to Spanish text translator pre-trained model is from [Helsinki-NLP/opus-
|
38 |
+
mt-en-es](https://huggingface.co/Helsinki-NLP/opus-mt-en-es) which is part of the <b>The Tatoeba Translation Challenge
|
39 |
+
(v2021-08-07)</b> as seen from its github repo at
|
40 |
+
[Helsinki-NLP/Tatoeba-Challenge](https://github.com/Helsinki-NLP/Tatoeba-Challenge). This project aims to develop machine
|
41 |
+
translation in real-world
|
42 |
cases for many languages. </p>
|
43 |
<h3>Text to Speech</h3>
|
44 |
+
<p style='text-align: justify'> The TTS model used is from [facebook/tts_transformer-es-css10]
|
45 |
+
(https://huggingface.co/facebook/tts_transformer-es-css10).
|
46 |
This model uses the <b>Fairseq(-py)</b> sequence modeling toolkit for speech synthesis, in this case, specifically TTS
|
47 |
+
for Spanish. More information can be seen on their git at [speech_synthesis]
|
48 |
+
(https://github.com/pytorch/fairseq/tree/main/examples/speech_synthesis). </p>
|
49 |
"""
|
50 |
|
51 |
|