Acoustics ML & DSP Engineer at Apple

Job description

As an Acoustics ML and DSP Engineer, you will develop new speech enhancement algorithms that combine machine learning and DSP, and take them from prototype to production on Apple devices. You will use and extend existing large-scale data preprocessing, simulation and cloud training pipelines, building new ones where needed. You will tune, optimize and listen to algorithms, and debug issues by tracing them to the data, the model or the signal chain. Your work will improve voice quality and intelligibility for telephony and voice calls in challenging conditions such as echo, double talk and noisy environments, and you will measure the effect of front-end processing on telephony quality and machine listening applications. You will also follow the state of the art in speech processing and adjacent fields, and work cross-functionally with tuning, DSP and audio ML colleagues within the team and with partners across the hardware and software organizations. MS with 3+ years of relevant experience, or PhD, in Acoustics, Audio, Electrical Engineering, Computer Science or a related field. Solid grasp of acoustics and signal processing fundamentals. Experience shipping audio or speech enhancement ML algorithms from prototype to production. Proficiency with PyTorch and Python, including version control, testing and reproducible experiments. Working knowledge of Max/MSP for real-time audio prototyping. Experience with causal, streaming models under real-time latency and compute constraints. Experience using, and where needed building, large-scale data and cloud training pipelines, including preprocessing and data simulation. Working knowledge of speech quality evaluation: objective metrics, telephony quality and echo performance measures, and subjective listening tests. Understanding of the state of the art in speech processing and adjacent fields. Enthusiasm for exploring and testing new ideas, with the ability to prioritize and deliver across several projects in parallel in a fast-paced environment. Team player with strong written and verbal communication, and the ability to work in cross-functional teams. Experience with audio DSP, such as echo and double-talk handling, microphone array processing, beam-forming, dynamics processing, EQ and noise suppression. Experience with hybrid systems that combine classical signal processing and machine learning. Familiarity with telephony voice quality requirements. Knowledge of front-end processing for speech recognition and machine listening. Experience with multi-modal machine learning. Strong coding and debugging skills (Python and C/C++), plus MATLAB for analysis. Knowledge of transducers and their acoustic integration. Curious and eager to try new ideas quickly, while keeping multiple product programs on track.

Requirements

Must-haves

MS with 3+ years of relevant experience, or PhD, in Acoustics, Audio, Electrical Engineering, Computer Science or a related field.
Solid grasp of acoustics and signal processing fundamentals.
Experience shipping audio or speech enhancement ML algorithms from prototype to production.
Proficiency with PyTorch and Python, including version control, testing and reproducible experiments.
Working knowledge of Max/MSP for real-time audio prototyping.
Experience with causal, streaming models under real-time latency and compute constraints.
Experience using, and where needed building, large-scale data and cloud training pipelines, including preprocessing and data simulation.
Working knowledge of speech quality evaluation: objective metrics, telephony quality and echo performance measures, and subjective listening tests.
Understanding of the state of the art in speech processing and adjacent fields.
Enthusiasm for exploring and testing new ideas, with the ability to prioritize and deliver across several projects in parallel in a fast-paced environment.
Team player with strong written and verbal communication, and the ability to work in cross-functional teams.
Experience with audio DSP, such as echo and double-talk handling, microphone array processing, beam-forming, dynamics processing, EQ and noise suppression.
Experience with hybrid systems that combine classical signal processing and machine learning.
Familiarity with telephony voice quality requirements.
Knowledge of front-end processing for speech recognition and machine listening.
Experience with multi-modal machine learning.
Strong coding and debugging skills (Python and C/C++), plus MATLAB for analysis.
Knowledge of transducers and their acoustic integration.
Curious and eager to try new ideas quickly, while keeping multiple product programs on track.

Nice-to-haves

None flagged.

Application prep

Use the role’s language and application details while tailoring your materials.

Keywords to mirror

mlmachine learningpython

Platform: apple
Effort: medium