Role
Develop machine-learning solutions for speech and audio applications, including ASR, TTS, speech enhancement, speaker-related tasks, and audio understanding.
Responsibilities
Develop and fine-tune speech recognition and audio models.
Prepare, clean, segment, and validate large-scale audio datasets.
Develop audio preprocessing and augmentation pipelines.
Evaluate models using metrics such as WER and CER.
Perform detailed error analysis across accents, environments, speakers, and recording conditions.
Fine-tune models for domain-specific speech such as call-center audio.
Improve model accuracy, latency, and robustness.
Work with VAD, diarization, speech enhancement, and related audio technologies.
Optimize models for production inference.
Requirements
Strong Python experience.
Strong understanding of deep learning and speech/audio processing.
Hands-on experience with ASR or other speech ML systems.
Experience with model training and fine-tuning.
Experience preparing and evaluating speech datasets.
Understanding of metrics such as WER/CER.
Strong debugging and error-analysis skills.
Nice to have
Experience with Persian speech recognition.
Experience with VAD, diarization, TTS, or speech enhancement.
Experience with audio processing tools such as FFmpeg.
Experience with Ray Serve, vLLM, or other inference frameworks.