Urgent-Voice recording project- Hmong, Cebuano, Luxembourgish, Tamazight, Tibetan, Turkmen, Kabyle, Ilocano, Balochi, Sepedi, Guarani, Madurese, Antillean Creole, Bouyei, Jamaican Creole, Sardinian, Rangpuri, Hiligaynon, and Bhili

We have a very interesting Voice recording project in Hmong, Cebuano, Luxembourgish, Tamazight, Tibetan, Turkmen, Kabyle, Ilocano, Balochi, Sepedi, Guarani, Madurese, Antillean Creole, Bouyei, Jamaican Creole, Sardinian, Rangpuri, Hiligaynon, and Bhili

. Here are the details. Can be done along with your family members, friends, colleagues. 

Task Description

You must record natural human conversational audio between two speakers in your assigned language to a total of 45 minutes (not per speaker). The same pair of speakers may participate in multiple recordings.

The recorded audio can be, either as:

• One single recording of 45 minutes, or
• Multiple shorter recordings
• Minimum length: 10 minutes
• Maximum length: 45 minutes

Recorded audio must be uploaded in MP3 format

• If your recording is not already in MP3 format, you may convert it using any free online audio conversion tool such as https://cloudconvert.com/ or https://www.freeconvert.com

When recording, ensure the following:

• The audio is mono (do not use stereo)
• The sample rate is at least 16 kHz
  • Smartphones/laptops usually already meet these requirements

To meet the total recording time of 45 minutes, you may record:

• • One file, or
  • Up to six files (adding up to 45 minutes)

Voice Style

Conversations should sound natural and realistic. You can include:

• Natural turn-taking
• Interruptions
• Overlapping speech
• Laughter
• Hesitations/filler words
• Varied vocabulary
• Natural code-switching
• Different speaking speeds and emotions

Please avoid

• Robotic or obvious reading
• Long silent pauses
• Excessive background noise
• Distorted or clipped audio
• Software, AI, or any audio source other than human voices
• Copyrighted media in the background (TV/music/radio)

Metadata

When uploading, the platform will require you to choose an option or fill in the space for certain aspects:

• Device type used for recording
• Topic/domain of the conversation
• Speaker first names
• Recording environment
• Recording type (spontaneous, scripted, or semi-scripted)
  • Scripted conversations are allowed only if they sound natural.
• Background noise level

If you are interested to participate in this project, the first step will be to register to our portal. The job link to upload the files can only be accessed by registered users.

Please Register on our platform using the following link:
https://freelancer.datamundi.space/register