At Papercup, we have developed a speech translation pipeline that has unlocked content from companies like Sky News and Discovery for hundreds of millions of viewers in their native language. As a PhD intern in the Machine Learning team, you will have the opportunity to further advance our industry leading speech synthesis technology by pursuing ambitious research collaboratively with the team. We have a track record of publishing at top tier speech processing conferences (INTERSPEECH 2020 and 2021 alongside a number of recent submissions to other conferences awaiting a decision), and you will join us in our effort to continuously push the boundaries of research in speech synthesis.
About the role:
We are looking for a curious, creative and intellectually bold self-starter who is eager to redefine what is possible in speech synthesis. You will be fully integrated within the 10-person ML team and participate in team planning processes, research discussions and knowledge sharing sessions. As part of this team, you will work with people who are familiar with both research and deployment of our multi-speaker, multi-lingual text-to-speech system. You will also have the opportunity to consult and collaborate with our esteemed science advisor, Simon King, who works with us day-to-day. There will be ample opportunity to get exposure to the product and interact with the humans-in-the-loop of our system, whether to obtain direct native speaker feedback or to deepen your understanding of the remaining gaps in TTS systems.
In this role, you will be pursuing TTS research collaboratively with other researchers in the team. You will own one or more research projects, conceptualise experiments, identify suitable datasets (public, proprietary or custom experimental commissions are all options), implement experiments and conduct evaluations. You will have access to dedicated compute resources, our data and evaluation infrastructure, and our research and deployment codebase to support your research. You are highly encouraged to submit a paper for publication in a top-tier machine learning conference. And of course, you’ll also be part of all company and team social events!
We will be accepting applications until Friday 15th April 2023, anywhere on earth. To apply, submit your CV. If your research interests align with the team’s, we will invite you for an interview and ask you to submit a research proposal or statement of research interest (not more than 750 – 1,000 words, approximately 2 pages) describing the project you would like to undertake. We currently have place for only one intern which we will fill on a rolling basis, so we highly encourage you to apply early.
You can find out more about our research on our engineering blog and samples page.
What you'll do with us:
- Do research on a topic of your choice within speech synthesis
- Conceptualise, implement and evaluate experiments in close collaboration with other researchers on the team
- Present research findings and developments, both verbally and in writing
- Work with your supervisor and the team to contribute to published research at a peer-reviewed conference and socialise your research through a blogpost
- Currently enrolled in a PhD programme (2nd year onwards) in machine learning or speech technology
- Available for 16-20 weeks (option to extend) within 2023, with start dates from April onwards
- Intellectually curious and eager to learn
- Able to formulate research ideas and translate them into concrete projects
- Proactive, organised and able to execute on ideas through completion
- Able to work collaboratively with other researchers and be an active participant in team research discussions
- Excellent critical thinking skills, with the ability to communicate ideas and thoughts clearly and convincingly, in both verbal and written form
- Familiarity with Python and experience working with machine learning frameworks (i.e. PyTorch/TensorFlow)
- Familiarity with machine learning theory - able to understand relevant prior research, comfortable discussing tradeoffs between different architectures and algorithms
Would be nice if you have:
- Have a track record of contributing to research, including paper publication in peer-reviewed journals or conferences, contribution to open source projects etc.
- Familiarity with speech synthesis, generative modelling or representation learning
Apart from all this good stuff, what else do we offer?
- Competitive salary of £65,000 pro rata
- Unlimited vacation policy
- Hybrid working: flex between WFH and time in our Old Street office (around 2 days per week)
- Private medical cover or monthly wellness bonus
- Learning budget and 'reading week' to carve out time to up skill in your domain
- The usual food and fun perks: snacks, beer fridge regular team socials, annual offsites
Please note we’re not looking for someone who ticks all the boxes, if you have some of the skills listed above and are willing to learn, you’re the person for Papercup.
The FlexScore® is the result of a rigorous 2-step verification of a company’s flexibility
First we assess the flexibility options Papercup provides and then we anonymously survey a statistically significant proportion of their employees to make sure Papercup is as flexible as they say they are. Our assessment is based on the six key elements of flexibility: location, hours, autonomy, benefits, role modelling and work-life balance.
We ask the hard questions so you don’t have to.
Working at Papercup
Gender diversity (male:female)
London - Aldgate
$20m Series A