-
I'm trying to learn a regression problem. The data is mostly one-hot encoded categorical variables, one continuous. The target output is a probability (0-1). Here is the code:
```
def read_lines(f…
-
参考该链接方法：
https://github.com/wenet-e2e/wenet/pull/1982
先尝试用作者提供训好模型、配置文件以及热词列表解码，在librispeech上test other wer为9.5%。（作者给出的结果为7.93%）
然后自己训练了一版，batch size 调整到32，以及batch num contex也调整到32, 迭代了29个ep…
-
时间戳结束时间大部分停在语音片段上，而不是静音处，如图：
使用模型： damo/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch
调用方式： modelscope pipeline
测试音频：[test-funasr.wav.zip](https://github.com/ali…
-
The `install_models.sh` file downloads 3 files, one is the `blstm.ep7.9langs-v1.bpej20k.model.py` file and the other two are `ep7.9langs-v1.bpej20k.bin.9xx` & `ep7.9langs-v1.bpej20k.codes.9xx`. `mlenc…
-
您好 想請問會release model?
我正在實現這篇論文，使用Voice Bank + DEMAND 資料集訓練但出來結果相差很多...
另外論文當中這部分有點奇怪，在equation (2)所表示的是elementwise adding，但在圖上最後卻是concat，這部分是依照哪個為主?
如果您願意回覆，非常感謝
-
Hi there,
Currently I had an idea of `Word-Level training`, basically it focuses on word-level recognition rather than the line-level & character-level that Tesseract is currently using.
The idea…
ghost updated
10 months ago
-
OS: Windows 10
Python/C++ Version：python 3.8.17
Package Version：pytorch==1.11.0、torchaudio==0.11.0、modelscope==1.8.1、funasr==0.7.3（pip list）
Model：damo/speech_paraformer-large-vad-punc_asr_nat-zh-c…
-
I f you want to use a pre trained `Transformer` for the same task, how would you use it instead of `LSTM` here? For example I want to use a lightweight `BERT` model, what would ne the changes to the l…
-
Hi everyone!
I ran the recipe given for ST on Must-C V1 (en-de) with the train_rnn configuration. The ACC goes up to 56% on the train after 14 epochs, but the validation BLEU remains at 0 (see scre…
-
We have been developing (enhancing) the ESPnet2 for the speaker diarization task. The [base code](https://github.com/espnet/espnet/pull/2939) follows the [EEND](https://github.com/hitachi-speech/EEND)…