menu
keyboard_arrow_up 3 keyboard_arrow_down

Startup Idea: Voice/Accent changer for call centers

Posted by mark

My idea is simple: I want to build an AI tool that allows live voice conversion to be used in call centers. The voice and accent from both ways would be converted in to a local accent to facilitate comprehension between the speakers (both ways).

The technical side of this idea is a whole challenge on its own and there are many things to consider like if we should also localize expressions and the words the people are using.

There's one company that seems to be doing something similar but my approach will be different especially in terms of marketing.

Bot application

Ultimately the goal is for the customer to first talk to a bot first that can take care of answering the most frequent questions. If the bot feels like the customer needs to talk to a human, then they can transfer to a representative.

The representative's voice will be the same as the bot so the customer won't even notice that they were transferred from a bot to a human...

I think this can really be beneficial in optimizing the operations of call centers.

Survey Results

Do you think this idea is worth pursuing?

10 answers

There will be a backlash to concept of "localizing" the voice of people?

Strongly disagreeDisagreeNeither agree nor disagreeAgreeStrongly agree

What pricing model do you think I should incorporate?

10 answers
addReview this post

Reviews (10)

keyboard_arrow_up 1 keyboard_arrow_down

Do you think this idea is worth pursuing?

Technically this is going to be tough. Your MVP might be very poor and not as good as you think especially in real world scenarios, running the process in real time, bad quality audio... So many issues, it would take a lot of investment just to make an MVP. Then you need to get customers which is a whole other problem. I do think however that this kind of technology is going to happen, it's just a question of time, you might be too early though?

There will be a backlash to concept of "localizing" the voice of people?

Neither agree nor disagree

What pricing model do you think I should incorporate?

The pricing model could be structured in several ways, depending on your target audience and value proposition. Options to consider include: - Subscription model: Offering tiered subscription plans for call centers based on usage and features. - Pay-per-minute: Charging call centers based on the minutes of voice conversion used. - Custom pricing: Tailoring pricing to the specific needs and scale of each call center. Also, consider offering a free trial or a limited-feature version to attract initial users and build a customer base.

A few remarks:

Technical aspect: Make sure you do not underestimate the difficulty of achieving something like this. You also probably need to integrate your system to existing call center systems. (Maybe this could be your target customer actually, not the call centers themselves, but the companies that build call center tech).

Marketing: It's a B2B startup which I believe makes it easier to market. You just need to convince the call center companies to give a try to your system. They do however need to trust you as the information that will go through your system can be sensitive. Maybe having a self-hosted solution that they can manage on their own would solve this problem.

Review by ybouane
keyboard_arrow_up 2 keyboard_arrow_down

Do you think this idea is worth pursuing?

For me, I'm pretty comfortable with bots answering my phone call as long as I get the answer I need, short and crisp. So that might not resolve my pain points

There will be a backlash to concept of "localizing" the voice of people?

Neither agree nor disagree

What pricing model do you think I should incorporate?

Pay as you go

I like the idea of applying AI into call center operations. I wonder how it will make the automated answers better and more customized without human input of customizing it?

Review by meeppupo
I see... Basically based on your review, I feel like I've been focusing too much on the tech "AI voice/accent changing" instead of trying to solve the problem which is that people need their questions answers / make changes through the phone. In the ideal scenario it should all be managed through AI. So the focus should be there, not trying to change the accent of the call representative.  — mark keyboard_arrow_up 1 keyboard_arrow_down

Do you think this idea is worth pursuing?

If you have the technical know-how and know that this is an actual problem. Why would a call center make use of this software? What benefit do they get?

There will be a backlash to concept of "localizing" the voice of people?

Disagree

What pricing model do you think I should incorporate?

Per minute, as the computation costs are not insignificant and depend on usage.

Have you actually talked with some call centres to see if they would actually need this?

Review by kingkhalid
I know that call centers train their representatives to speak more like the people they serve and it's a big deal for them. But I haven't talk to any yet, so for sure I should do some market research before investing time and effort in the technical aspect of this problem.  — mark keyboard_arrow_up 1 keyboard_arrow_down

Do you think this idea is worth pursuing?

This feels like a hard problem, unless your an expert in this area. If you feel capable of this kind of development then it's possible you may be able to use your skill more effectively on a smaller project. That said, I imagine if it is done right, companies would pay huge sums for the software. Also be aware that a similar idea has been put out there before - https://www.sanas.ai/

There will be a backlash to concept of "localizing" the voice of people?

Neither agree nor disagree

What pricing model do you think I should incorporate?

I think you would need feedback from your actual customer to establish a pricing point and model. I'd be guessing and wildly wrong I think.

See above

Review by benjamin
keyboard_arrow_up 1 keyboard_arrow_down

Do you think this idea is worth pursuing?

yes

There will be a backlash to concept of "localizing" the voice of people?

Disagree

What pricing model do you think I should incorporate?

price per minute/conversation, with a cap

Do just one thing (technically):

  • software that can perfectly understand indian accent in realtime (speech to text), it needs to run on commodity hardware in under 0.5s
    if you have that, text to speech is relatively solved

Sell just one thing: plug in my box/software in the audio path, i eliminate the accent from your employees speech in realtime

you don't need to do the other way around.

this means the companies can hire anybody with good conversation skills that do not have a "perfect English/American accent" (which is the hard part)

Review by ruslan_t
keyboard_arrow_up 0 keyboard_arrow_down

Do you think this idea is worth pursuing?

I don't think there needs to be any more clandestine services than there already is. This feels ethically wrong.

There will be a backlash to concept of "localizing" the voice of people?

Strongly agree

What pricing model do you think I should incorporate?

It would make sense to use a usage metric for pricing. Eg. 0.006c per action

Overall, I don't like the idea. It feels ambiguously dishonest, especially when I feel we should be moving towards improved authenticity.

Review by rentables
keyboard_arrow_up 1 keyboard_arrow_down

Do you think this idea is worth pursuing?

yes. seems pretty cool. I think the way of doing this would be to rapidly expand, get a good product.

There will be a backlash to concept of "localizing" the voice of people?

Neither agree nor disagree

What pricing model do you think I should incorporate?

I think the model should be try and get people using it and then sell it to a big company.

I like the idea a lot. although it does have chalenges

  • finding accents
  • figuring out which accent matches to which customer
  • applying accent in real time, any delay would be bad
Review by yepster
Yup there are several technical challenges with this idea and latency/delay is one of the major ones!  — ybouane keyboard_arrow_up 1 keyboard_arrow_down

Do you think this idea is worth pursuing?

Yes, I think its amazing.

There will be a backlash to concept of "localizing" the voice of people?

Strongly agree

What pricing model do you think I should incorporate?

pricing should be based on the localizing or region.

Make a landing page with X account. create content around it and check if is it worth it or not. You can create an idea video to share or create hype.

Review by htmlmonk
keyboard_arrow_up 0 keyboard_arrow_down

Do you think this idea is worth pursuing?

I would be careful in trying not to reinforce stereotypical accents of cultures in your program. Other than that, this sounds like a great idea.

There will be a backlash to concept of "localizing" the voice of people?

Strongly disagree

What pricing model do you think I should incorporate?

I think you should set it at $20 a season.

I like this idea.

Review by igobysoandso
keyboard_arrow_up 0 keyboard_arrow_down

Do you think this idea is worth pursuing?

yes

There will be a backlash to concept of "localizing" the voice of people?

Strongly disagree

What pricing model do you think I should incorporate?

usage hours should be the way to charge.

will be abused (most probably by the india call centers) but, logically it might work.

Review by hasbehas