How To Create Podcast In Your Own Voice Using AI
- January 27, 2018
- Posted by: admin
- Category: Technology
Let’s talk about audio content today, let’s talk about a place from where we can get a lot of traffic, a source many people are not using very effectively and that source is Podcasts. First, we will talk about what a podcast is and why there is some insane number behind one. Please do share our post with your friends if you like them.
Last week we did an experiment with audio content but it was a failed experiment as of now since technology is in a very early stage but even after a failed experiment I want you to know about this amazing technology because it is going to be a great force in future.
What is podcast?
The term Podcast was derived in 2004 by Ben Hammersley in an article of The Guardian known as Audible revolution. Now if we look at the term we can break it down in ‘Pod’ as in iPod and ‘Cast’ as in ‘Broadcast‘. In simple term, it’s an online radio which is more than a traditional radio because it’s kind of open source and anyone can podcast instead of only RJs of a radio station.
Note – Just because the term comes from iPod it doesn’t mean only iPods have podcasts, it is a multi-channel entity and spread across devices and operating systems.
How To Start A Podcast?
I think this life hacker article beautifully explains things required to start a Podcast and how to start one.
What do Numbers say About Podcasts?
- About 60% people in the USA know what podcast means, the estimated number of people familiar with podcast is 168 million in the USA
- Around 112 Million people in USA listen to Podcast
- Podcast spend have been massively increased with nearly 250 million
- Unlike video, only 5% do not consume podcast and massively 85% consume entire episode
- Podcast already have a huge audience
- There is massive opportunity for growth
- You can create a brand name there and reach more and more people
How To Create Audio of Yourself without Even Speaking a Word?
Problem statement – You don’t have much time to record tons of audio content, maybe you are a founder of a company, travelling a lot and physically cannot be present to record audio or you are ill. What if you give your digital voice to someone so that they can type the word and it creates a digital audio in your voice from that text itself, it looks like a fairy tale right?
Then let us tell you how you can do it
Why Did We mention This As A Failed Experiment At The Starting Of this Blog?
There are various reason for that, one of them is the limitation of the tool which lets you record audio of 300 characters only which roughly is 20 seconds, also there is a click between sounds so you need to slow the pace of the audio via an audio editor like Audacity.
By recording your own voice you will have better quality, faster recording the only bottleneck in this is that you can only record for 20 seconds and then you have to edit it.
What Do You Need?
- A good mic – If you do not want to spend more than use this mic, it will cost you only INR 199. here is amazon link to the same
- An editing software –Audacity – Its Free and open source
- A close room where noise is minimum
- Account in Lyrebird Artificial Intelligence Programme – Its Free but have limitation
Let’s Start With The Resources We have!
Before starting this experiment we want to let you know this technology can be easily misused by creating fake news kind of content which promote yellow journalism, so we are not endorsing any of the bad implementations and this failed experiment is only for educational purposes.
We need to find a quiet place where we can record our voice and attach our mic to our laptop to record sample voice
As you can see it mentioned Only one minute of audio. In reality, you can record maximum 300 characters.
Click on Create Your Digital Voice, you will land on a sign-up page
Once you signup they will ask you to verify email
Verify Email and log in, once you log in you will ask for record sample audio, click ‘Start Now’
Once you click ‘ Start Now!’ they will give you some instruction, as shown in the picture, in short, you need to
- Speak Loudly
- Speak Clearly
- Record at least 30 Voice sample ( In our case we recorded 60)
Now the next screen is recording screen, record the text written in bold, speak slowly so that it can record your proper voice but not too much slowly
In our case, we recorded “But much of the challenge of writing code is editing the code you or someone else has already written.”
Allow use of microphone in your browser too if it asks for it
Once you record you will see the option of Listening, validate and recording again. Listen to your sample and if you find it clear enough then validate or record again.
Once you validate it will go through verification process
Once they system validate the audio sample you will see change in 30 recording count
Now you need to keep recording your audio sample until it says we have enough sample of yours
Once the Artificial Intelligence learn your enough voice data they will let you know
Click on ‘Create my digital voice’, you will land up on a page where you can put your text and generate your audio content
We are going to put content from this blog post and generate a voice sample from Arbab Usmani’s voice
In a second you will get your audio recording
This is the original recording – https://lyrebird.ai/g/ISsm7Za4
Now since this technology is a very early stage we are getting some clicks of voice, so we have to remove it by slowing it down and adding some royalty free music in the track.
Launch Audacity and import your audio in it
Now we are not a professional sound editor we are using whatever knowledge we have to enhance things, if you have a better solution then please let us know at our Facebook Group
Now click on effect and choose Normalize
You will land in a window where normalize amplitude of the audio, we found – 15 dB is where clicking is not audible, you can choose your own number after listening to preview
Now lets add some music behind it to fill up more gap, I suggest you to go to cc0 Audio to avoid any copyright strike, I used this, just import your music in another track and mix them as shown in image.
After mixing export them as Mp3
Now, this is in Audacity you can merge multiple audio files to make a long version of your audio story, We have edited just one. to edit multiple just add another audio in the same track and you will have it
Here is final sample –
PS – We Know the voice quality is not that great and background sound is annoying to some people, that is why we call it a failed experiment.