Offer summary

Qualifications:

Native proficiency in Taiwanese Hokkien or Amoy Hokkien, Minimum of 2 years experience in language evaluation or related field, Familiarity with audio response evaluation and feedback processes, Access to a personal laptop or PC for remote work..

Key responsabilities:

Create audio prompts and provide feedback on the model's audio responses.

Evaluate and improve the model's output focusing on accent and fluency.

Iterate and generate multiple audio responses to achieve high-quality ratings.

Aim to identify 2 to 5 audio responses with a perfect rating of 7 on accent.

Job description

Greetings from Summa Linguae Technologies Pvt Ltd!

We have multiple short term assignment for Taiwanese Hokkien OR Amoy Hokkien Language (Fujian Province). Please find below Job description:

Role: Taiwanese Hokkien OR Amoy Hokkien Language Expert

Language proficiency: Natives

Experience: 2+ years /5 years of experience in language

Duration: 20 - 80 hours / 2 - 3 weeks (may extend depending on the project)

Scope of work: The Audio Prompt Response Evaluation for Accent and Fluency aims to help an LLM application to improve fluency, accent, and dialect of their model audio responses.

Location: Remote

Interview Process: 90 mins of online assessment

Employment: Consultant of Summa Linguae Technologies Pvt Ltd

Payment Terms: 30 - 45 days from the receipt of invoice

Job Description:

Project Brief:

The goal is to assess the model's audio responses for a specific language/accent combination and help it learn/improve/fine-tune its accent/fluency in the target language. Participants are expected to give audio prompts, generate audio responses, rate on certain parameters including accent/fluency and keep on iterating/generating audio responses till at least 2 perfect responses are achieved.

Roles and responsibility:

Will be interact with the model in a designated language/accents by creating audio prompts and providing feedback on the model’s audio responses.
Your goal is to evaluate and improve the model’s output from the accent/fluency’s POV. Content of the prompt or the response doesn’t matter.
The goal is to keep on evaluating/iterating till you find at least 2 and a maximum of 5 responses with a perfect rating of 7 on accent. To get to this, you might have to generate any number of responses (10, 20, 30, etc.)
This means you should find between 2 and 5 audio responses that sound the best, with a score of 7 being the highest possible rating for how natural and fluent the model's speech is from the target language/locale’s accent POV .

Skills required:

Must have at least 2+ years of experience in Chinese (Meixian dialect) Language

Must have their own Laptop/PC etc.

If interested, please share your CV and expected remuneration to Recruitment_Apac@summalinguae.com

Required profile