方言を話すおしゃべり猫型ロボット『ミーア』をリリースしました(こちらをクリック)

[Mia] The result of an open call for dialect narration. Utilization of Coconara, Crowdworks, and Lancers

mia-dialect-narration
この記事は約18分で読めます。

Introduction.

I am developing “Mia,” a talking cat-shaped robot that speaks various dialects.

When it was on display in the storefront at Tsutaya Plus, a customer who visited

I think it would be interesting to have a “Tsugaru-ben” dialect. Beforehand, when I was checking out the Tsugaru Plus exhibit, my kids were glued to it. The cat models are cute.

We received feedback from them. We felt that there was a need to expand the number of dialects, so we decided to greatly expand the number of dialects supported this time.

Try public solicitation (Crowdworks, Lancers, Coconara, our own website)

For the first four dialects we released (Osaka dialect, Hiroshima dialect, Kagoshima dialect, and Hakata dialect), we used a crowdsourcing service called Coconara to find and request voice actors.

However, this time we wanted to support the dialects of 43 other prefectures all at once, so it was very difficult to find and send a message to each voice actor one by one.

Moreover, although some voice actors’ profiles include sample voices for various characters, ages, and genders, there are few cases where the dialects they can handle are clearly stated, probably because there is not much need for them. Therefore, even if a voice actor was actually capable of handling a particular dialect, there was a concern that if we were to search for them, we would not be able to find them due to a lack of information, and we would lose out on the opportunity to request their services.

So, why don’t we put conditions on the application and invite the public to apply? We decided to write a letter of invitation.

In the case of narration, many voice actors present their conditions at the unit price per character plus the unit price per file cut, and in this case, we set the amount slightly higher than the market price, respectively, as shown below.

3 yen per character
10 yen per cut

Click here to see the actual recruitment letter we created.

Thank you very much for taking a look at our request.

Summary]
We are a company developing a talking cat-shaped robot “Mia” that speaks dialect.
Here are the details of our services.

ミーア
おしゃべり猫型ロボット「ミーア」| 子どもへのクリスマスプレゼントに

For this request, we are looking for someone to record the original narration of the dialects spoken by the cats.
Please see below for videos of Mia speaking each dialect.
Currently, Osaka dialect, Hakata dialect, Hiroshima dialect, and Kagoshima dialect have already been recorded.

Requests ]
The amount of words: less than 3,000 (about 2,700-3,000 characters).
Number of cuts: 150 phrases (about 20 characters per phrase). 150 phrases with about 20 characters per phrase.
Number of applicants: 1 person per prefecture (*Please see below for details)
Materials to be provided: Scenario manuscript (Excel)
Delivery format: 16bit 44.1Khz monaural wav format
Publication of results: Yes (If you wish, we will give you a credit in the dialect section of the application you operate)
Experience: Beginners OK (Experienced persons are welcome)
Retakes: Up to one.
File naming convention: 1.wav, 2.wav, etc. sequential numbers (phrase ID.wav)

Charge】】】 1 character 3 yen
3 yen per character
(We would like you to start and stop recording each line with a smart phone.)
The basic fee is as above, but can be negotiated.

We are looking for people like this.
⭐️ If you can speak the dialects of the following prefectures (except for the strike-through lines, which are confirmed requests), please contact us. (Dialects for which requests have been confirmed are strikethrough).

Must be able to respond quickly.
Please contact us if you are interested in our services.
Hokkaido, Aomori (Tsugaru dialect), Iwate, Miyagi, Akita, Yamagata, Fukushima, Ibaraki, Tochigi, Gunma, Saitama, Chiba (Boshu dialect), Tokyo, Kanagawa, Niigata,Toyama, Ishikawa, Fukui, Yamanashi (Koshu dialect), Nagano, Gifu (Mino dialect), Shizuoka, Aichi (Nagoya dialect, Mikawa dialect), Mie, ShigaKyoto, Osaka, Hyogo (Kobe dialect), Nara, Wakayama, Tottori, Shimane (Izumo dialect), Okayama, Hiroshima, Yamaguchi, Tokushima, Kagawa, Ehime, Kochi (Tosa dialect), Fukuoka, Saga, Nagasaki, Kumamoto, Oita, Miyazaki, Kagoshima, Okinawa
oooooooooooooooooooooooooooooooooooooooooooooooooo

Delivery date and method
Within 10 days from the delivery of the scenario draft.

Requests when applying
Dialects in which you can narrate (Okinawa dialect, Okayama dialect, Tochigi dialect, etc…)
If you have a portfolio of past work that shows your voice and narration quality, please provide it.

At first, we posted the openings only on Coconara, but then we thought, “If it’s the same openings, wouldn’t we get more people if we posted them on other crowdsourcing sites and on our own site?” So, I posted a copy and paste of the recruitment text on CrowdWorks and Lancers, as well as on Mia’s website.

https://mia-cat.com/notice/mia-dialects-narration

An unexpectedly large number of dialects were gathered.

I thought I would be satisfied if I could get about 5 dialects, especially the Tsugaru, Okinawan, and Kyoto dialects, which I personally wanted, but right after the call, I received quite a few applications. I was so surprised.

Finally, the following number of people were recruited via each site.

  • Crowdworks: 41
  • Lancers: 13
  • Coconara: 23 items
  • Mia sites: 3

As expected, the largest number of respondents came via CrowdWorks, the largest crowdsourcing service.

I was surprised to see that there was also a google form directly from the recruitment page posted on the Mia website. I thought it would not be caught in a search since we are not that well known yet.

Items Focused on in Selection

We received more applications than we had imagined, so we wondered what the selection criteria would be and decided to focus on the following two points

  • Does the tone of the voice fit Mia’s image?
  • Is the communication exchange going to be smooth?

I thought it was pretty obvious at the time of application, especially with regard to the second, communication. As I went from the top of the application below to the bottom, I decided that the person was able to anticipate our intentions, and decided to give priority to this request.

  • Only one word, “I’m applying” (no mention of which dialects you can support)
  • Only copied and pasted the profile text (did not say which dialects they can support, probably did not read the application carefully).
  • Profile text + supported dialects in text only -> Hard to judge since there are no sample voices for dialects
  • Supported dialects are attached as sample voices
  • Look at Mia’s website and SNS in advance, create your own corresponding dialect of phrases that Mia is likely to speak and attach it as a sample voice.

Busy working on scripts.

Each dialect has 150 phrases and requires a script for each dialect.

Creating the dialect script was quite a painstaking and tedious process, involving Googling “Saga dialect list,” “Saga dialect example sentences,” etc., and listening to people speaking the dialect on YouTube and TikTok to create the content.

I also used the “Love Dialect Conversion” site and ChatGPT. However, it was difficult to convert dialects in one shot as I wanted, perhaps because the database of dialects is small.

https://www.8toch.net/translate

I created a sheet for each dialect in google spreadsheet, and wrote the dialect and its meaning in the sheet. It took about 2-3 hours per dialect. This time, I created scripts for more than 20 dialects at once.

The prepared script was sent to the narrator with the following message

We have created this document while looking at information on the Internet, so if there are any phrases in the Okayama dialect that you feel uncomfortable with, you may change them as you see fit and speak accordingly.

The target age group is from 20s to 60s. If there are phrases that are only used by the elderly (70s and older), could you please correct them as well?

If so, we would appreciate it if you could include a copy of the revised script with the audio recording.
Thank you in advance.

Glad to hear from the narrator.

Although we did not expect it at all at the beginning, we received some pleasant words from the voice actors who narrated the program, so we would like to introduce some of them.

We are very grateful for your quick response and the smooth recording process!
Mia is very cute and I would like to purchase her personally when she is ready!

I am following you here on X. If it is implemented, may I also announce it as an appearance report on SNS?

I visited your website and was impressed with the background of Mia development.

I volunteered at a nursing home when I was a student and I truly believe that communication is powerful for people.
I am happy to be involved in the creation of Mia.

I will do my best to bring a smile to as many people’s faces as possible.
I look forward to receiving your script. Thank you very much.

Thank you for your assistance.

Thank you very much for taking the time to check in with us!
We are very happy to be involved in a job that you wanted to do so badly.

I enjoyed it so much that I couldn’t help grinning at some of the local jokes in the dialogue!
I’m planning to visit Tokyo in September, so I’d love to go to the store in Marunouchi and see the actual product (^^).

Thank you very much for your time.
Please let us know if there is anything else we can do for you.

We had no idea that they would be pleased with the work we had commissioned them to do.

The audio we received was so good that I wanted to listen to it over and over again, and there were some dialects that I actually kept repeating. Professional voice actors may have the right recording equipment and environment, and there may be no noise, but I felt from listening to the actual recorded voice that they are professionals in putting emotion into a single phrase.
I also realized that there is still a difference between an amateur and a professional, and that it is better to ask a professional voice actor or announcer to do the job.

In fact, after taking the public offering style this time, we realized that the public offering style is more convenient, including management, when we have to respond to a large number of requests at once with services already in place here.

We have added just under 20 dialects at once and plan to release them in early September!

タイトルとURLをコピーしました