jyleong / voiceminder-server

Personal profile website made with Hugo

Commands

Make new post:

> hugo new blog/file.md
> hugo new code/file.md

Run locally:

> hugo server -D

Compile assets to target location:

> hugo -d ../(directory)

voiceminder-server's Issues

Countdown on Raspberry fails

Threading on raspberry is way too slow. With speech recognition, socket shuts down unexpectly. To reproduce error, run speech_Client.py

ACTIONS
In the pi branch, i will remove countdown class to see if it improves performance

Client Speech Recognition

When Clients speaks, it should not be listening. When it is listening, it should not be speaking

Approach 1 : Theading mutex lock on speaking Boolean

Approach 2 : Investigating use of Observable pattern

Make websocket-client demo run again

also document the commands needed to run demo

For Countdown class, have way to terminate/exit the single thread

Do this instead of closing all threads that are non main

Or use processes and know how to close specific process

Make countdown the same for both users conversing

idea is to have countdown keep track of both users
current implementation is a separate countdown per users
possible implementation idea is to have a conversation class between two users and have the countdown class check that.

record useful commands into one document

Feature Request :Only Respond if the first word is greeting

"Hey Kevin, pick up some cheese" should return success/fail

case success: say nothing.

case fail: say "Kevin's not available"

Dataflow:
Extract name
look up socket by name
if no socket by that name, fail.
send to socket
if socket unresponsive, fail.

Refactor, make websocket.py less heavy

websocket.py is getting dense
are there ways we can decouple some of its code

voiceminder-client does not work on mobile

likely due to no https, lacking openssl.

Needs research for iOS webRTC

TODO:
-Android Chrome - should be possible, look to see how others have made it work.

isaffirmative in processText

Server Crashes when user tries to converse with himself

solution: dont let user talk to themselves!

remove threading name staging

no more

is your name ???

server error on web when conversing with 3 people.

needs research.

user.conversant.conversant is a potential point of failure

may also be a failure with 2 people.

-works with text with 3 people, but not with voice.

"My name is June" should return "Hello June"

executable for testing

processText one word response crash

fix

Message Types

'I am June' is a different type and than 'hi june can you send me some cheese'
Use python enum to differentiate different types

UserList Unit Test

blank message on text client crashes the client

server remains just fine, can re-register as same user if you re-boot the client.

TechDebt: Web Speech LifeCycle

Access property names of websocket to understand how to properly use it

Socket status 200

to reproduce error:
run:
python3 src/run.py
lt -s voiceminder -p 5000
vim echo_client.py
change to websocket.create_connection("ws://voiceminder.localtunnel.me")
python3 echo_client.py

on client's side, error shown:
--- request header ---
GET / HTTP/1.1
Upgrade: websocket
Connection: Upgrade
Host: voiceminder.localtunnel.me
Origin: http://voiceminder.localtunnel.me
Sec-WebSocket-Key: 56Z81kA/Uw6oy92EUTDJSQ==
Sec-WebSocket-Version: 13

---

--- response header ---
HTTP/1.1 200 OK
Server: nginx/1.10.1

Date: Mon, 20 Nov 2017 14:31:15 GMT
Content-Type: text/html; charset=utf-8
Content-Length: 13
Connection: keep-alive

Traceback (most recent call last):
File "echo_client.py", line 6, in
ws = websocket.create_connection("ws://voiceminder.localtunnel.me")
File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/websocket/_core.py", line 487, in create_connection
websock.connect(url, **options)
File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/websocket/_core.py", line 214, in connect
self.handshake_response = handshake(self.sock, *addrs, **options)
File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/websocket/_handshake.py", line 69, in handshake
status, resp = _get_resp_headers(sock)
File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/websocket/_handshake.py", line 129, in _get_resp_headers
raise WebSocketBadStatusException("Handshake status %d", status)
websocket._exceptions.WebSocketBadStatusException: Handshake status 200

Run tornado websocket demo

handle conversing state

Timer in different class

to a separate class. Then import it, incase we will use our countdown in other parts of voiceminder

confirm name in websocket.py

callback: timer on user thread and expire timer as soon as they respond.

client times out after ~5 seconds of inactivity, cannot be revived.

on the client side, listening state will stop recognizing microphone input after ~5 seconds, but can still be revived if messaged by someone else.

Suggested resolves:
-event handlers
-onspeechend
-onaudioend

make repo private

tune listen performance

Function "socket by name"

Conversation object

easier to reason about start and end of conversation
problems observed

-it is echoing sometimes
-seems like it gets confused with user.name and recipient name
-3way is shaky at best

separate listen ands speak state on client

Create Database for Managing User and Conversation

Create the following tables:
User(uid, username, nickname, m_at, c_at)
Conversation(user1(FK User), user2(FK User), cid, m_at, c_at)
Conv0_Msgs(cid(FK Conversation), uid(FK User), msg, m_at, c_at)

for james, create tables
for jon and kai, investigate tornado db access