About The Opportunity
A fast-scaling AI innovations firm in the Speech & Voice Technology sector, we build and validate next-gen voice AI systems that power intelligent assistants, call-center automation, and multilingual speech recognition for global enterprises. Our team ensures voice models understand accents, intonations, and edge-case utterances across Indian and global languages—delivering human-grade accuracy at scale for mission-critical applications.
Role & Responsibilities
- Evaluate speech recognition, speaker verification, and voice synthesis models using structured rubrics across noise conditions, accents, and domains.
- Annotate, transcribe, and flag audio samples for model training, focusing on phonetic accuracy, intent detection, and emotional tone.
- Collaborate with AI engineers to triage model failures, surface bias patterns, and recommend data augmentations for low-performing utterances.
- Perform comparative testing of voice AI outputs across vendors or model versions, documenting performance deltas in standardized dashboards.
- Maintain quality control benchmarks and contribute to continuous improvement of evaluation protocols for voice AI systems.
- Support internationalization efforts by testing multi-accent and code-switching capabilities in Indian and global language pairs.
Skills & Qualifications
Must-Have
- Speech transcription
- Audio annotation tools
- Phonetic transcription
- Accent evaluation
- Speech recognition metrics (WER, CER)
- Quality assurance for AI models
- Microsoft Excel
- Google Forms
Preferred
- Experience with Praat or Audacity
- Knowledge of IPA (International Phonetic Alphabet)
- Background in linguistics or phonetics
Benefits & Culture Highlights
- Work with cutting-edge voice AI models deployed across Fortune 500 call centers and consumer devices.
- On-site collaborative environment with weekly syncs, tech talks, and hands-on access to speech pipelines.
- Impact-driven role where your evaluations directly improve model accuracy for millions of users.
Skills: transcription,edge,audacity,annotation,models,access,testing,speech,assistants,speech recognition
