Comments (4)
I have fixed it, because all the training voice is about -6db, but the voice in our device maybe about -12db, -18db, current model weights can not cover this case, I retraining the model with -18db,0db random scale, and it worked.
from nnom.
The log info:
./rnn_denoise fip11c_16k.wav fip11c_out_16k.wav
Model version: 0.4.3
NNoM version 0.4.3
To disable logs, please void the marco 'NNOM_LOG(...)' in 'nnom_port.h'.
Data format: Channel last (HWC)
Start compiling model...
Layer(#) Activation output shape ops(MAC) mem(in, out, buf) mem blk lifetime
#1 Input - - ( 1, 1, 40,) ( 40, 40, 0) 1 - - - - - - -
#2 RNN/GRU - - ( 1, 96, ) 86k ( 40, 96, 1808) 1 1 1 - - - - -
#3 RNN/GRU - - ( 1, 96, ) 102k ( 96, 96, 1920) 1 1 1 - - - - -
#4 RNN/GRU - - ( 1, 48, ) 32k ( 96, 48, 1056) 1 1 1 - - - - -
#5 Flatten - - ( 48, ) ( 48, 48, 0) - - 1 - - - - -
#6 Dense - HrdSigd - ( 20, ) 960 ( 48, 20, 96) 1 1 1 - - - - -
#7 Output - - ( 20, ) ( 20, 20, 0) - 1 - - - - - -
Memory cost by each block:
blk_0:1920 blk_1:1808 blk_2:96 blk_3:0 blk_4:0 blk_5:0 blk_6:0 blk_7:0
Memory cost by network buffers: 3824 bytes
Total memory occupied: 7512 bytes
Processing file: fip11c_16k.wav
Print layer input/output..
Layer(#) - Input(Qnm) Output(Qnm) Oshape
#1 Input - 0. 7 0. 7 ( 1, 1, 40,)
#2 RNN/GRU - 0. 7 0. 7 ( 1, 96, )
#3 RNN/GRU - 0. 7 0. 7 ( 1, 96, )
#4 RNN/GRU - 0. 7 0. 7 ( 1, 48, )
#5 Flatten - 0. 7 0. 7 ( 48, )
#6 Dense - 0. 7 0. 7 ( 20, )
#7 Output - 0. 7 0. 7 ( 20, )
Print running stat..
Layer(#) - Time(us) ops(MACs) ops/us
#1 Input - 0
#2 RNN/GRU - 0 86k
#3 RNN/GRU - 0 102k
#4 RNN/GRU - 0 32k
#5 Flatten - 0
#6 Dense - 0 960
#7 Output - 0
Summary:
Total ops (MAC): 222480(0.22M)
Prediction time :0us
Total memory:7512
Noisy signal 'fip11c_16k.wav' has been de-noised by NNoM.
The output is saved to 'fip11c_out_16k.wav'.
from nnom.
when I run the prebuild demo, there is an error:
./rnndenoise fip11c_16k.wav fip11c_16k_out.wav
Model version: 0.4.3
Please set static memory using 'nnom_set_static_buf()' before calling model created.
from nnom.
when I run the prebuild demo, there is an error: ./rnndenoise fip11c_16k.wav fip11c_16k_out.wav Model version: 0.4.3 Please set static memory using 'nnom_set_static_buf()' before calling model created.
How to fix it? Can you tell me? Thank you very much
from nnom.
Related Issues (20)
- about reshape HOT 3
- 使用了per channel量化和kld量化方法后,出现了多次推理结果不一致的问题
- 关于输出维度的问题 HOT 4
- nnom静态内存支持如何打开? HOT 6
- main_pc.c中的test_x.txt和text_y.txt如何制作?
- 关于识别几秒时长的语音 HOT 3
- inhomogeneous shape after 1 dimensions
- keyword_spotting中的main_pc.c中的test_x.txt和text_y.txt如何制作?
- nothing
- 关于使用DW_Conv2D与Conv2D的移植后使用耗时的问题 HOT 1
- 使用#define NNOM_USING_CMSIS_NN 的错误。 HOT 2
- 运行rnnnoise的example报错 HOT 2
- Error: Deprecated Usage of `np.int` in `to_cstyle` Function in `gen_config.py`
- Keras 3 version compatibility HOT 1
- Installable using pip
- Exception with Tensorflow 2.16 HOT 2
- Consultation on kws examples?
- RNN state makes model output tend towards 0 HOT 2
- kws HOT 1
- RISCV rv32g undefined reference to `__divdi3'
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from nnom.