Expert Audio Transcriber (Urdu)

About the Job

Perle is an AI infrastructure leader that partners with global pioneers like Samsung, xAI, and Unisys to develop expert-driven training data and evaluation systems. The company is currently seeking Expert Urdu Transcribers to convert native-speaker audio recordings into precise, conventionally formatted written transcripts. This role transcends standard data entry, requiring high-level linguistic judgment to shape the quality of AI models through accurate orthography, speaker attribution, and the handling of complex code-switching.

Transcribers will work with a variety of audio, ranging from clean studio environments to natural, real-world conversational speech, primarily focusing on Standard Urdu as spoken in major urban centers like Karachi, Lahore, Islamabad, Delhi, Lucknow, and Hyderabad.

Core Responsibilities

• Linguistic Transcription: Converts audio into time-aligned text while strictly adhering to project-specific style guides.
• Detailed Annotation: Applies precise conventions for overlapping speech, speaker identification, and non-speech events such as laughter or background noise.
• Orthographic Decision-Making: Makes principled choices regarding dialectal nuances and Urdu-English script-mixing.
• Quality Control: Performs rigorous self-reviews of all submissions and collaborates with the QA team on calibration sessions and style guide updates.
• Issue Reporting: Identifies and flags ambiguous segments or audio quality concerns using established project protocols.

Candidate Requirements

Required Qualifications:

• Native Fluency: Must possess native-level command of Urdu with lifelong immersion in Pakistan or Urdu-speaking regions of India.
• Script Mastery: Demonstrated high accuracy in reading and writing the Nastaliq script.
• Professional Experience: At least two years of experience in transcription, translation, journalism, or linguistic annotation.
• Technical Setup: Access to a quiet workspace, professional-grade headphones, and high-speed internet.
• Precision: Proven ability to maintain word-level accuracy while meeting strict deadlines.

Preferred Assets:

• An academic background in Urdu linguistics, literature, or journalism.
• A working knowledge of Hindi to assist with cross-lingual vocabulary or shared audio contexts.
• Familiarity with Phonetic systems (IPA) or specific transcription conventions like "clean-verbatim."

Engagement & Compensation

• Evaluation: New hires begin with a paid pilot task to ensure alignment with project standards.
• Structure: Compensation is provided on a per-task or per-minute basis, with rates reflecting the complexity of the audio.
• Flexibility: The role offers flexible scheduling, with the ability to scale volume based on active client projects.
• Growth: Successful contributors gain a direct relationship with linguistic leads and may be invited to participate in additional expert domains.

Application Process

Interested candidates must apply via the Winnow Portal. The process involves setting up a profile and completing an automated interview lasting approximately 25 minutes. Applications are typically reviewed within 24 hours, and successful candidates are moved immediately into the project onboarding phase.