Research Scientist – Speech AI

Company: Meeami Technologies Private Limited Location: Hyderabad, Telangana

Experience: 4+ Years

Qualification: Bachelor’s / Master’s / PhD in AI/ML, CSE, ECE, or related field

About Meeami Technologies

Meeami Technologies is a leader in Audio AI, Noise Cancellation, Speaker Identification, and Audio Analytics. With 20+ years of experience and 75+ granted and pending patents, Meeami develops advanced audio, voice, video, and communication technologies for OEMs, semiconductor companies, call centers, carriers, and other customers.

Meeami Technologies Website

Role Overview

We are looking for a Research Scientist – Speech AI with 4+ years of hands-on experience in Speech, Audio, or Voice AI. The role involves a combination of AI research and production engineering, including developing, evaluating, optimizing, and deploying speech/audio models. You will formulate research problems, design experiments, analyze model performance, reproduce SOTA research, and take solutions from prototype to production.

Candidates can have expertise in one or more of the following areas:

1. Automatic Speech Recognition (ASR)

- Develop end-to-end ASR systems using Transformer, Conformer, RNN-T, CTC and related architectures.
- Work with Whisper, HuBERT, wav2vec 2.0, WavLM and speech foundation/self-supervised models.
- Develop streaming and low-latency speech recognition systems.
- Work on noisy, multi-accent and low-resource speech conditions.
- Experience with Indian-English or regional Indian accents is highly relevant.
- Research accent-invariant representations and improve cross-speaker/accent generalization.
- Evaluate models using WER, CER, RTF, latency and memory.

2. Text-to-Speech (TTS)

- Develop modern TTS systems using Diffusion / Flow Matching and related architectures.
- Experience with VITS, CosyVoice, ZipVoice or similar SOTA approaches is preferred.




- Work on forced alignment, MFA, MAS, duration and time-aligned TTS.
- Research accent adaptation, speaker representations, prosody and style conditioning.
- Work on voice cloning, speaker similarity, naturalness and intelligibility.
- Optimize TTS systems for real-time and low-compute inference.
- Evaluate using MOS, speaker similarity, intelligibility, prosody and RTF.

3. Speech Enhancement / Noise Suppression

- Develop speech enhancement models in time-domain, spectral-domain and hybrid approaches.
- Work with single-channel and multi-channel audio.
- Experience with beamforming is an advantage.
- Improve speech quality under low-SNR, reverberation and non-stationary noise conditions.
- Optimize models for real-time and edge/embedded deployment.
- Experience with SI-SDR, PESQ, STOI and DNSMOS is preferred.

4. Model Optimization

- Work on quantization, Quantization-Aware Training (QAT), pruning and knowledge distillation.
- Optimize speech/audio models for edge and embedded deployment.
- Experience with ONNX, TensorRT or similar inference optimization tools.
- Profile models across CPU/GPU/NPU platforms.
- Optimize latency, throughput, memory and compute requirements.
- C/C++ and CUDA experience is preferred for this track.

Key Skills & Qualifications

Must Have:

- 4+ years of hands-on experience in Speech AI / Audio AI / Voice AI.
- Strong programming skills in Python.
- Hands-on experience with PyTorch and/or TensorFlow.
- Strong understanding of Machine Learning and Deep Learning.
- Experience with CNNs, RNNs, LSTMs and Transformers applied to speech/audio.
- Robust fundamentals in Digital Signal Processing, Linear Algebra, Optimization, Statistics and Pattern Recognition.
- Ability to independently design experiments, analyze results and debug models.
- Strong research mindset and problem-solving skills.
- Hands-on understanding of Generative AI, including GANs, Diffusion Models and Flow Matching.

📌 Meeami- Research Scientist(Audio/Speech) (Hyderabad)
🏢 NextHire
📍 Hyderabad