https://github.com/crodriguez1a/deepspeech-stt
A slim Python client for Mozilla's DeepSpeech STT
https://github.com/crodriguez1a/deepspeech-stt
deep-speech mozilla python speech-to-text
Last synced: 3 months ago
JSON representation
A slim Python client for Mozilla's DeepSpeech STT
- Host: GitHub
- URL: https://github.com/crodriguez1a/deepspeech-stt
- Owner: crodriguez1a
- Created: 2020-07-12T19:03:23.000Z (almost 5 years ago)
- Default Branch: master
- Last Pushed: 2020-07-24T00:57:42.000Z (almost 5 years ago)
- Last Synced: 2025-02-09T01:33:46.503Z (5 months ago)
- Topics: deep-speech, mozilla, python, speech-to-text
- Language: Jupyter Notebook
- Homepage:
- Size: 20.1 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# deepspeech-stt
## Introduction
A slim Python client for Mozilla's [DeepSpeech](https://github.com/mozilla/DeepSpeech/) speech-to-text
## Usage
```python
from src.deepspeech_stt import deepspeech_predict
ouput_text: str = deepspeech_predict(
wav_file_path,
batch_after_silence=True,
silence_threshold=45, # 45db
filters=["logmmse_denoise", "butter_bandpass_filter"]
)
```
Parameter | Default | Description
---|---|---
`wave_filename` | `None` | Path to wave file
`batch_after_silence`|`True`| Create batch from input splitting after natural gaps of silence
`silence_threshold` | `50` | The threshold (in decibels) below
reference to consider as silence
`filters`| `None` | List of signal filters to apply as pre-processing:
`butter_bandpass_filter`, `high_pass_filter`, `low_pass_filter`, `logmmse_denoise`
See [notebook](notebooks/Examples.ipynb) for examples
## Installation
Download [Mozilla's DeepSpeech 0.7.4](https://github.com/mozilla/DeepSpeech/releases) pre-trained model (~200mb)
Then run:
```
poetry install
poetry shell
```