Speech Processing Python Packages

silero-vad

Silero VAD: pre-trained enterprise-grade Voice Activity Detector

whisper-timestamped

Multilingual Automatic Speech Recognition with word-level timestamps and confidence

pysptk

A python wrapper for Speech Signal Processing Toolkit (SPTK).

swift-f0

Fast and accurate fundamental frequency (F0) detector using convolutional neural networks

nnmnkwii

Library to build speech synthesis systems designed for easy and fast prototyping.

indic-num2words

Python library for converting numbers to words for all Indian Languages.

spafe

:sound: spafe: Simplified Python Audio Features Extraction

rutextnorm

Fast Russian Text normalization for TTS using only RegEx.

silero-vad-lite

Lightweight wrapper for Silero VAD using internal ONNX Runtime and with no python package dependencies

crisperwhisper

Controllable Transcription. Verbatim ( every, filler, pause, stutter, vocal sound) , or intended ( what the speaker meant to say, optimized for readability) with word-level timestamps.

resemble-enhance

AI powered speech denoising and enhancement

torchscale

Foundation Architecture for (M)LLMs

voicefixer

General Speech Restoration

fastwer

A PyPI package for fast word/character error rate (WER/CER) calculation

diarize

Speaker diarization for Python — "who spoke when?" CPU-only, no API keys, Apache 2.0. ~10.8% DER on VoxConverse, 8x faster than real-time.

premove-itn

Premove ITN: context-aware inverse text normalization for English voice-agent transcripts.

dpdfnet

Clean up noisy speech in real time with DPDFNet - open-source streaming speech enhancement for research, audio apps, and edge devices. Includes pretrained models, PyTorch code, ONNX/TFLite inference, 8/16/48 kHz support, and live demos.

voicetut-tts

VoiceTut-TTS is an Egyptian-Arabic text-to-speech system fine-tuned from OmniVoice on ~380 hours of Egyptian podcast speech. It produces natural Egyptian speech with seamless Arabic ↔ English code-switching, ships 15 built-in studio voices, supports zero-shot voice cloning

bournemouth-forced-aligner

Extract phoneme-level timestamps from speeh audio.

everyvoice

The EveryVoice TTS Toolkit - Text To Speech for your language

attenlabs-saa

Addressee detection for voice agents: device-directed speech detection that runs before STT, so background speech, side conversations, and the agent's own TTS echo never trigger it. No wake word, model-agnostic, drop-in for LiveKit, Pipecat, ElevenLabs, Twilio, and OpenAI. The layer your VAD and turn detection are missing.

stark-engine

S.T.A.R.K. - Speech And Text Algorithmic Recognition Kit

funtalk

语音处理统一封装：ASR 基于 Whisper 做音频转写，TTS 支持 Azure 与 Edge TTS，提供一致的基类接口便于切换后端

scoreq

SCOREQ: Speech COntrastive REgression for Quality Assessment (NeurIPS 2024)