TalkToPod combines AI powered podcast transcripts with ChatGPT to enable users to chat with their favorite podcasts.

TalkToPod

AI turns your favorite podcasts into a chat bot

TalkToPod

As a podcast fanatic, I personally am interested in connecting the dots between stories and trying to find specific niche details buried in podcasts. I built TalkToPod so I can simply ask a question and access the relevant information in the form of an AI-generated answer.

My strongest need was semantic search over podcasts, but it's quicker and more fun when the information is extracted into an AI generated answer with citations.

How it works

Podcasts are downloaded and the audio stream is segmented by pseudonymous speaker (diarisation).

Each segment is transcribed using OpenAI's open source Whisper model.

Pseudonymous podcast speakers are then identified one of three ways:

1) The top of the transcript is passed to ChatGPT's API to infer speakers by the text of the transcript and pattern of speakers (determining if speakers introduce each other or refer to each other by name in a pattern that can be used to identify their pseudonymous tags)

2) Unidentified speakers can be matched to previously previously vectorized voice prints

3) Manual identification

The transcripts with the identified speakers are embedded and stored in a vector database.

When a user asks a question, semantic search is used to find relevant sections of transcripts (the question is embedded and cosine similarity is used to find related portions of transcripts).

The question and related portions of transcripts are passed to ChatGPT to determine the answer to the question citing the appropriate transcripts.

Examples:

ExpressAI

AI-Powered Tailored Messaging for Recruiting Top Talent

ExpressAI

The pain of attracting top-tier talent is most acute for startup founders. Your startup might not have widespread name recognition, but ExpressAI lets you stand out by doing what startups do best - being creative and getting in front of people who matter. With ExpressAI your outreach messages stand out in the LinkedIn Inmail inboxes of top talent.

I use a unit file to run Airflow Scheduler as a daemon that should restart any time it fails. But I noticed the airflow scheduler doesn't restart when it can't connect to the backend (scheduler) database.

I'm using MariaDB (like MySQL) as the backend (scheduler) database, but it's on a separate VM. So if that VM starts after Airflow Scheduler, then it's stuck in an unhealthy state and doesn't restart.

Here's the MySQL error in the Airflow Scheduler

● airflow-scheduler.service - Airflow scheduler daemon
     Loaded: loaded (/etc/systemd/system/airflow-scheduler.service; enabled; vendor preset: enabled)
     Active: active (running) since Sun 2023-02-12 00:28:21 EST; 1 day 9h ago
   Main PID: 152 (airflow)
      Tasks: 2 (limit: 134815)
     Memory: 71.0M
        CPU: 7.994s
     CGroup: /system.slice/airflow-scheduler.service
             ├─152 /usr/bin/python3 /home/airflow/.local/bin/airflow scheduler
             └─420 airflow serve-logs

Feb 12 00:28:29 anonAirflow airflow[152]:   File "/home/airflow/.local/lib/python3.8/site-packages/mysql/connector/__init__.py", lin>
Feb 12 00:28:29 anonAirflow airflow[152]:     return CMySQLConnection(*args, **kwargs)
Feb 12 00:28:29 anonAirflow airflow[152]:   File "/home/airflow/.local/lib/python3.8/site-packages/mysql/connector/connection_cext.p>
Feb 12 00:28:29 anonAirflow airflow[152]:     self.connect(**kwargs)
Feb 12 00:28:29 anonAirflow airflow[152]:   File "/home/airflow/.local/lib/python3.8/site-packages/mysql/connector/abstracts.py", li>
Feb 12 00:28:29 anonAirflow airflow[152]:     self._open_connection()
Feb 12 00:28:29 anonAirflow airflow[152]:   File "/home/airflow/.local/lib/python3.8/site-packages/mysql/connector/connection_cext.p>
Feb 12 00:28:29 anonAirflow airflow[152]:     raise errors.get_mysql_exception(msg=exc.msg, errno=exc.errno,
Feb 12 00:28:29 anonAirflow airflow[152]: sqlalchemy.exc.DatabaseError: (mysql.connector.errors.DatabaseError) 2003 (HY000): Can't c>
Feb 12 00:28:29 anonAirflow airflow[152]: (Background on this error at: http://sqlalche.me/e/13/4xp6)
root@anonAirflow:/# sudo systemctl status airflow-scheduler.service | grep sqlalchemy.exc.DatabaseError

Restart Airflow Scheduler

So I added a cronjob to test every hour if the scheduler was health, and if it was unhealthy to restart!

58 * * * * if [ "$(curl http://1.2.3.4:8080/health | jq .scheduler.status)" != '"healthy"' ]; then systemctl restart airflow-scheduler.service; fi

And for anyone looking for the unit file I'm using for the scheduler, please see below!

#
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements.  See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership.  The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License.  You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied.  See the License for the
# specific language governing permissions and limitations
# under the License.

[Unit]
Description=Airflow scheduler daemon
After=network.target postgresql.service mysql.service redis.service rabbitmq-server.service
Wants=postgresql.service mysql.service redis.service rabbitmq-server.service

[Service]
EnvironmentFile=/etc/sysconfig/airflow
User=airflow
Group=airflow
Type=simple
ExecStart=/bin/airflow scheduler
Restart=always
RestartSec=5s

[Install]
WantedBy=multi-user.target

airflow/airflow-scheduler.service at main · apache/airflow

Apache Airflow - A platform to programmatically author, schedule, and monitor workflows - airflow/airflow-scheduler.service at main · apache/airflow

GitHubapache

Given the age of Ubuntu 20.04 vs Ubuntu 22.04, installing python3.10 in Ubuntu 20.04 requires reading from a PPA rather than the regular software packages. That means to upgrade python to 3.10 on Ubuntu 20.04 you'll need to add a community repository.

How to Add Package Python 3.10 to Ubuntu 20.04

Commands below will require sudo - but I'm not including it so that you don't copy and paste sudo commands!!!

apt install software-properties-common

add-apt-repository ppa:deadsnakes/ppa

apt-get update

How To Install Python 3.10 After Adding Package

apt-get install python3.10 python3.10-distutils python3-mysqldb python3.10-dev

Word of Caution: you want to be careful when adding a PPA of developers you don't know. So why should you trust deadsnakes? Well, use at your own risk, but it's been run by Anthony Sottile and Felix Krull for years and is the most widely recommended way to install new versions of python.
https://askubuntu.com/questions/1398568/installing-python-who-is-deadsnakes-and-why-should-i-trust-them

It had been a couple months since updating Arch (bad, I know, but it's headless and I forgot!)

When I went to run an update I saw this error happen...

sudo pacman -Syu
:: Synchronizing package databases...
 core                                                                      998.0   B  1399   B/s 00:01 [#############################################################] 100%
 extra                                                                     998.0   B  37.5 KiB/s 00:00 [#############################################################] 100%
 community                                                                 998.0   B  37.5 KiB/s 00:00 [#############################################################] 100%
error: GPGME error: No data
error: GPGME error: No data
error: GPGME error: No data
error: failed to synchronize all databases (invalid or corrupted database (PGP signature))

So I inspected the database - and it was an HTML page!

less /var/lib/pacman/sync/core.db

The .db file clearly should not be a webpage - it turns out the mirror url was stale. So I  moved the entire sync directory...

sudo mv /var/lib/pacman/sync /var/lib/pacman/sync.BROKEN

And updated the mirror I was using by commenting out the broken one

> sudo nano /etc/pacman.d/mirrorlist

# I commented out the first 
# Server = https://archlinux.qern-industries.pw/$repo/os/$arch
Server = https://america.mirror.pkgbuild.com/$repo/os/$arch
Server = https://mirror.hackingand.coffee/arch/$repo/os/$arch

Ran the update again - but more things were also stale >.<

(61/61) checking package integrity                                                                     [#############################################################] 100%
error: openssl: signature from "Pierre Schmitz <pierre@archlinux.org>" is marginal trust
:: File /var/cache/pacman/pkg/openssl-3.0.7-4-x86_64.pkg.tar.zst is corrupted (invalid or corrupted package (PGP signature)).
Do you want to delete it? [Y/n] y
error: failed to commit transaction (invalid or corrupted package)
Errors occurred, no packages were upgraded.

# Fix keys
sudo pacman-key --init
sudo pacman-key --populate archlinux

# Update keyring
sudo pacman -Sy archlinux-keyring

# Update system
sudo pacman -Syu

(Solution for keyring found https://wiki.archlinux.org/title/Pacman/Package_signing#Upgrade_system_regularly)

So just make these simple bash scripts (the first to run long tests and the second to print out the smartctl stats, make them executable with chmod +x smartctl_all_test.sh and run them!

Here's a script for quickly starting self-tests for many drives at once:

#!/bin/bash
for drive in `ls /dev/sd? /dev/nvme0n?`; do
  echo $drive;
  smartctl -t long $drive;
done

And another to see the most critical indications of failure (in my opinion)

#!/bin/bash
for drive in `ls /dev/sd? /dev/nvme0n?`; do
  echo $drive;
  smartctl -a $drive | grep 'Reallocated_Sector_Ct\|Reallocated_Event_Count\|Offline_Uncorrectable\|Error_Rate\|Power_On_Hours\|Wear_Leveling_Count\|Used_Rsvd_Blk_Cnt_Tot\|Available Spare\|Percentage Used\|Media and Data Integrity Errors';
  echo; echo;
done

Disclaimer:

At some point (perhaps mid-2021), AWS seems to have disabled Spot as an option for P-class instances. It may be possible to request a limit increase to run a P spot instance. See notes on instance types at the end.

Background

I had a web scraping job that collected a lot of news articles I wanted to summarize using machine learning (PEGASUS, to be specific). Summarization required a large EC2 instance that I didn't want to leave running - so I decided to "bid" on a spot instance every day when I needed it. Spot prices were often 25 cents on the dollar and only run when needed so overall my deep learning predictions were 98% cheaper than if on an on demand instance running all the time.

Below is a breakdown of how to request a spot instance and run a task on it for deep learning or whatever your use case may be! I used python/boto3 for this.

Python Initiated Spot Request

Import

Import packages and create client

import boto3
import base64
import datetime

client = boto3.client("ec2", "us-east-2")

User Data

Define the user data - this is what's executed once the instance boots

userData = '''
Content-Type: multipart/mixed; boundary="//"
MIME-Version: 1.0

--//
Content-Type: text/cloud-config; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment; filename="cloud-config.txt"

#cloud-config
cloud_final_modules:
- [scripts-user, always]

--//
Content-Type: text/x-shellscript; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment; filename="userdata.txt"

#!/bin/bash
sudo shutdown -P +120; aws s3 cp  s3://my-bucket-name/run_full_prediction_flow_on_ec2.sh ./ && sudo runuser -l ubuntu  -c '/bin/bash /tmp/run_full_prediction_flow_on_ec2.sh > run_pred.log 2>&1'
--//
'''

cloud-config controls when userdata runs (in this case userdata runs when your instance launches).

You'll see that userdata here is a bash script that:

• sets the shutdown time for 2 hours from launch (as a fail-safe to prevent cost overruns) sudo shutdown -P +120
• fetches a bash script aws s3 cp  s3://my-bucket-name/run_full_prediction_flow_on_ec2.sh ./
• runs that retrieved bash script sudo runuser -l ubuntu  -c '/bin/bash /tmp/run_full_prediction_flow_on_ec2.sh > run_pred.log 2>&1

Here, s3://my-bucket-name/run_full_prediction_flow_on_ec2.sh is some shell script I'm fetching and executing to instantiate anything the instance needs (package installations/configurations) and then running the prediction. An example of what this could look like at the end of this post for reference.

Instance Specification

Then I define the instance specification:

launchSpecification = {
        "IamInstanceProfile": {
            "Arn": "arn:aws:iam::MY-AWS-ACCOUNT-ID:instance-profile/MY-INSTANCE-IAM",
        },
        "BlockDeviceMappings": [
            {
                "DeviceName": "/dev/sda1", # note: '/dev/xvda' on some instance types
                "Ebs": {
                    "DeleteOnTermination": True,
                    "VolumeSize": 80,
                    "VolumeType": "standard",
                },
            },
        ],
        "ImageId": "ami-MY-AMI-ID",
        "KeyName": "MY-SSH-KEY",
        "SecurityGroups": ["MY-SG-1-NAME", "MY-SG-2-NAME"],
        "InstanceType": "m4.xlarge",
        "Placement": {
            "AvailabilityZone": "us-east-2a",
        },
        "SecurityGroupIds": ["sg-MY-SG-1", "sg-MY-SG-2"],
        "UserData": base64.b64encode(
            userData.encode(
                "utf-8"
            )
        ).decode("ascii"),
    }

Make request

And finally I make the request with the price I'm willing to pay

client.request_spot_instances(
    DryRun=False,
    SpotPrice="0.30",
    ClientToken=datetime.datetime.now().isoformat()[:10],
    InstanceCount=1,
    Type="one-time",
    LaunchSpecification=launchSpecification
    )

ClientToken is a unique string that AWS uses to ensure your request isn't submitted multiple times. datetime.datetime.now().isoformat()[:10], is the date which means that I can't accidentally spin up multiple instances in the same day.

Final script

import boto3
import base64
import datetime

client = boto3.client("ec2", "us-east-2")

userData = '''
Content-Type: multipart/mixed; boundary="//"
MIME-Version: 1.0

--//
Content-Type: text/cloud-config; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment; filename="cloud-config.txt"

#cloud-config
cloud_final_modules:
- [scripts-user, always]

--//
Content-Type: text/x-shellscript; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment; filename="userdata.txt"

#!/bin/bash
sudo shutdown -P +120; aws s3 cp  s3://my-bucket-name/run_full_prediction_flow_on_ec2.sh ./ && sudo runuser -l ubuntu  -c '/bin/bash /tmp/run_full_prediction_flow_on_ec2.sh > run_pred.log 2>&1'
--//
'''

launchSpecification = {
        "IamInstanceProfile": {
            "Arn": "arn:aws:iam::MY-AWS-ACCOUNT-ID:instance-profile/MY-INSTANCE-IAM",
        },
        "BlockDeviceMappings": [
            {
                "DeviceName": "/dev/sda1", # note: '/dev/xvda' on some instance types
                "Ebs": {
                    "DeleteOnTermination": True,
                    "VolumeSize": 80,
                    "VolumeType": "standard",
                },
            },
        ],
        "ImageId": "ami-MY-AMI-ID",
        "KeyName": "MY-SSH-KEY",
        "SecurityGroups": ["MY-SG-1-NAME", "MY-SG-2-NAME"],
        "InstanceType": "m4.xlarge",
        "Placement": {
            "AvailabilityZone": "us-east-2a",
        },
        "SecurityGroupIds": ["sg-MY-SG-1", "sg-MY-SG-2"],
        "UserData": base64.b64encode(
            userData.encode(
                "utf-8"
            )
        ).decode("ascii"),
    }


client.request_spot_instances(
    DryRun=False,
    SpotPrice="0.30",
    ClientToken=datetime.datetime.now().isoformat()[:10],
    InstanceCount=1,
    Type="one-time",
    LaunchSpecification=launchSpecification
    )

For Reference:

Instance type notes:

If you hit an error like botocore.exceptions.ClientError: An error occurred (MaxSpotInstanceCountExceeded) when calling the RequestSpotInstances operation: Max spot instance count exceeded check https://us-east-2.console.aws.amazon.com/ec2/v2/home?region=us-east-2#Limits: to see what instance types you're able to use.  When I searched for spot in June 2022 I saw that "All P Spot Instance Requests" only allowed "0 vCPUs" explaining why my old script with p2.xlarge was failing!

For deep learning you'll want a GPU instance.  AWS seems to have created a new class that has GPU's and can be run as spot - the DL instance type - but that's like $15+ per hour! You may be able to request a limit increase to run P instances if you need a GPU.

run_full_prediction_flow_on_ec2.sh

#cloud-boothook
#!/bin/bash
set -e
export LC_ALL=C.UTF-8
export LANG=C.UTF-8
PATH=/home/ubuntu/.local/bin:$PATH

# Set up directories
mkdir /home/ubuntu/analysis
mkdir /home/ubuntu/analysis/pred

# fetch the data from S3
aws s3 sync s3://MY_BUCKET/data  /home/ubuntu/analysis/data
aws s3 sync s3://MY_BUCKET/model  /home/ubuntu/analysis/model

cd /home/ubuntu/analysis/model

# Install packages
pip3 install --user -r requirements.txt
pip3 install --user pipenv ipython
pipenv --site-packages
pipenv install

# Predict
pipenv run python run_predictions.py

# Send predictions to S3
aws s3 sync /home/ubuntu/analysis/pred s3://MY_BUCKET/pred/`date +"%Y%m%d"` 

# Shutdown after making predictions to limit costs. This was commented out in case someone happened to copy and paste this accidentally...
# sudo shutdown

I had a website scraping project that scraped tens of thousands of websites every day. I was running this on an EC2 instance, but didn't want the EC2 instance's IP to be blocked by any of these websites.

One solution for scraping with (near) infinite IP's that I've come across is utilizing AWS API Gateway such as with requests-ip-rotator. But, with hundreds of thousands of requests a day, I was afraid the costs could add up so I used a VPN.  The VPN I was using would rotate through it's own IP's pretty regularly (sorry, but I'm going to be fuzzy on the exact details of what I was using). It wasn't an insane number of IP's like API Gateway would have been, but enough that I could rotate as needed.

Launching the VPN (openvpn) led to me losing access to everything in the VPC! I could no longer access my RDS instance - the VPN was routing all my traffic - including what should have stayed within the VPC!

I still needed to access my RDS instance inside of the VPC, so I routed that traffic specifically through the appropriate gateway.

route add -host  my-rds-instance.us-east-2.rds.amazonaws.com  gw 172.31.16.1

I had service that launched the VPN, so I configured this to run immediately after the VPN started.

There was a recent Axios article about how government benefits will be a drag going into 2022.

However, this ignores the fact that the largest increase in government transfers, the economic impact payments took place in Q1/Q2, and expanded unemployment benefits ran out in Q3.

Q4 experienced 6.9% quarter-over-quarter growth (SAAR) despite government social benefits to persons falling from 20.6% of personal income to 18.6% of personal income from Q3 into Q4.

Year-over-year there will be a decline in benefits, but the growth in compensation of employees more than offsets it - and the growth in Q4 alone shows how small the effects of the declining government benefits will be.

There was a recent Axios article about how government benefits will be a drag going into 2022.

However, this ignores the fact that the largest increase in government transfers, the economic impact payments took place in Q1/Q2, and expanded unemployment benefits ran out in Q3.

Q4 experienced 6.9% quarter-over-quarter growth (SAAR) despite government social benefits to persons falling from 20.6% of personal income to 18.6% of personal income from Q3 into Q4.

Year-over-year there will be a decline in benefits, but the growth in compensation of employees more than offsets it - and the growth in Q4 alone shows how small the effects of the declining government benefits will be.

For the indices below, we compare to December 2019, just before the pandemic.

Durable goods cost 12.0% more in December 2021 than they did in December 2019, while non-durable goods and services only cost 6.7 and 6.3 percent more, respectively (as shown above).

Looking at actual expenditures, consumers spent 29% more on durable goods in December 2021 relative to December 2019 (not only were durables more expensive, but consumers were buying more of them). Expenditure on services didn't grow much though (only 6% over two years) (as shown below).

But... if we think about how much people typically spend on durables relative to other goods we actually see a slightly different story.

While expenditure on durables grew 29%, personal income grew 12.4%.  Again, looking at percentages it looks bad, but expenditure on durables only grew from 8.2% of personal income to 9.5% of personal income.  

While that ate into budgets some, the rise in expenditure on services was less than the rise in income - and services are a much larger expenditure (51.3% of income in December 2021) that the personal savings rate actually rose from 7.3 in December 2019 to 7.9% in December 2021.

While inflation on it's surface looks like it would harm consumers, on average it hasn't actually been widespread enough to cut into savings compared to December 2019.

First - here's the problem and a vague allusion to the solution:

Ad Blockers Affect Mixpanel

Depending on your website and the type of ad blocker, some users who have ad blockers enabled can prevent Mixpanel from loading on your page.As a result, Mixpanel will not be able to collect data ...

Mixpanel Help Center

Okay, here's the code I had before the fix - this reaches out to Mixpanel directly and is therefore blocked by many Ad Blockers.

mixpanel.init(mixpanel_token);

So I add in the proxy and tell Mixpanel to ignore the IP address of my proxy server (otherwise all my users would have the same IP Address).

const proxy_url = 'https://myproxy.example.com/';

mixpanel.init(mixpanel_token, { api_host: `${proxy_url}https://api-js.mixpanel.com`, ip: 0 });

But there's one more piece we need to add! The actual IP address! This help page suggests you can pass the IP address so I figured I'd give it a shot.

Fetch the user's IP address (not shown) and register it so it's passed with every Mixpanel call within this session.

mixpanel.register({
      ip: ip,
});

Don't be creepy with this. Only use tracking to improve your platform, not to exploit users!

Self-hosting services on EC2 instances can come with unexpected behavior. If the service is hosted using Nginx, it typically expects a "Host" header. If you have Cloudfront between the EC2 instance and the internet (e.g., for caching, delivery, etc.), Cloudfront won't forward the "Host" header by default.

In Cloudfront settings, go to "Behavior."

Select "Use legacy cache settings"

Select "Whitelist" as the method to whitelist Headers

Select "Host" then click "Add >>"

Wait for Cloudfront to deplay and you should be good!

Additional resource:

Configure CloudFront to forward the Host header to the origin

Amazon Web Services, Inc.

Cloudfront Nginx Rewrite Causing Problems

I am setting up a cdn using cloudfront. My origin for my cloudfront distribution is my aws load balancer (ELB). When I make a request to cloudfront instead of getting the cloudfront url (cdn.mysite...

Stack Overflowgprime