About this role
Role Overview
- Research optimization solutions for terminology thesaurus from papers and design reasonable terminology filtering and hotword optimization solutions.
- Implement multi-language hotword algorithms based on SpeechLLM and optimize their effects; collaborate with the engineering team to deploy the hotword recognition solution.
- Combine scenario data to fine-tune the speech recognition model and improve ASR recognition effects across multiple languages and industries.
- Build a test set and system for keyword recognition and industry recognition engines, and evaluate the terminology recognition and industry engine effects of open-source models and commercial interfaces.
Requirements
- 3 to 5 years of speech algorithm training experience, with experience in fine-tuning and training SpeechLLM.
- Experience processing hundreds of thousands of hours of speech data and training speech recognition models.
- Familiar with SpeechLLM, speech SSL training, with from-scratch training experience for models similar to StepAudio, Qwen3omni, etc. Individual contributors responsible for model training within teams like the StepAudio speech group are preferred.
- Papers in top speech conferences like Interspeech, ICASSP, or patents related to speech.
Benefits
- Top-tier healthcare for employees and dependents, including dental and vision, and a generous employer subsidy.
- 401(k) plan for full time employees with company matching.
- Unlimited PTO, plus 13 paid holidays.
- 12 weeks of paid time off to spend time with your new family, regardless of gender.
- Minimum of 3x in office per week.
- New hires are equipped with their choice of new top-of-the-line laptops and workstation setups.
- Best office equipment. Annual offsites. Free office drinks and snacks.