Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
機械学習のための音声信号処理(基礎編)/ Speech Signal Processing
Search
moonlight-aska
October 21, 2018
0
1.3k
機械学習のための音声信号処理(基礎編)/ Speech Signal Processing
2018年10月21日開催の「大江橋Pythonの会#3」の資料です.
moonlight-aska
October 21, 2018
Tweet
Share
More Decks by moonlight-aska
See All by moonlight-aska
Create Your Own AI with Dify×Gemma3
aska
0
37
Generative AI Prototyping
aska
0
5
【入門】プロンプトの書き方のコツ / Tips for writing prompts
aska
0
190
CHATGPT。はじめの一歩 / ChatGPT. Get Started
aska
0
130
「Kingyo AI Navi」アプリ / Kingyo AI Navi App
aska
0
260
Kingo AI Navi LINEをもっと使い倒せ!!
aska
0
140
Depth画像で物体検知やってみたー。/ Objects Detection with Depth Images
aska
0
790
Kingyo AI Naviアプリ開発 / Kingyo AI Navi App
aska
0
430
AutoML Vision Edgeで金魚分類モデルを学習してみた / Kingyo Classification Model with AutoML Vision Edge
aska
0
570
Featured
See All Featured
Designing Experiences People Love
moore
143
24k
Exploring the Power of Turbo Streams & Action Cable | RailsConf2023
kevinliebholz
37
6.2k
It's Worth the Effort
3n
188
29k
Facilitating Awesome Meetings
lara
57
6.7k
技術選定の審美眼(2025年版) / Understanding the Spiral of Technologies 2025 edition
twada
PRO
115
100k
Designing for Timeless Needs
cassininazir
0
120
Typedesign – Prime Four
hannesfritz
42
2.9k
Collaborative Software Design: How to facilitate domain modelling decisions
baasie
0
120
Reality Check: Gamification 10 Years Later
codingconduct
0
2k
The Curious Case for Waylosing
cassininazir
0
210
Balancing Empowerment & Direction
lara
5
840
4 Signs Your Business is Dying
shpigford
187
22k
Transcript
None
NARA
안녕하세요
None
f P t
None
None
None
None
None
None
Convolutional Recurrent Neural Network
[音情報処理論 音声処理における信号処理1より引用]
None
None
×
×
[音声言語処理特論 第2回音声認識の基礎、DPマッチングの基礎より引用]
None
None
None
None
None
None
https://www.kaggle.com/c/tensorflow-speech-recognition-challenge
None
MFCC12 ΔMFCC12 ΔΔMFCC12
None
None
None
None
--- No. 6795 edc53350_nohash_0.wav (house ) --- * 1位 :
house (0.999820) 2位 : cat (0.000060) 3位 : off (0.000032) 4位 : yes (0.000024) 5位 : down (0.000013) --- No. 6796 e95c70e2_nohash_0.wav (house ) --- * 1位 : house (0.999953) 2位 : off (0.000012) 3位 : cat (0.000005) 4位 : eight (0.000004) 5位 : happy (0.000004) --- No. 6797 258f4559_nohash_0.wav (house ) --- * 1位 : house (0.999980) 2位 : off (0.000007) 3位 : happy (0.000004) 4位 : cat (0.000004) 5位 : eight (0.000002) --- No. 6798 1657c9fa_nohash_0.wav (house ) --- * 1位 : house (0.999972) 2位 : off (0.000011) 3位 : yes (0.000003) 4位 : happy (0.000003) 5位 : cat (0.000003) ---------- Total Accuracy ---------- 1位 : 93.57 % ( 6361 / 6798 ) 2位 : 96.78 % ( 6579 / 6798 ) 3位 : 97.78 % ( 6647 / 6798 ) 4位 : 98.35 % ( 6686 / 6798 ) 5位 : 98.57 % ( 6701 / 6798 )
None
NARA
None
None
None
https://ai.googleblog.com/2017/08/launching-speech-commands-dataset.html