EverySpeech by Anand (Piper TTS - Fast CPU)

WARNING: Piper TTS FAILED TO LOAD. Speech synthesis will not work. Check Space logs and `requirements.txt` for `piper-tts` and `onnxruntime`.

Fast CPU-based Text-to-Speech!

Write text in Input Text.
S1=(description) and {characteristic} tags are currently stripped before sending to Piper TTS.
Non-verbal tags like (laughs) or (coughs) will be read as text by Piper (e.g., it will say the word "laughs").
Background Music/Sound: Use [sound: name, volume: X] (e.g., [sound: rain, volume:0.3]). The first such tag applies to the whole output. Sound files from sounds directory. Optional Gemini Text Enhancement:
Enable, then provide a prompt like "Make this text more formal:". Your original text is appended.
Provide API key if not in Space secrets. Piper TTS Parameters: Control voice speed/style. Current voice: en_US-lessac-medium

Using Piper voice: en_US-lessac-medium. Adjust parameters for style.

Length Scale (Prosody/Speed)

Smaller is faster. Affects prosody.

0.5 2

Noise Scale (Variability)

Adds noise for expressiveness. Default: 0.667

0 1.5

Noise W (Phoneme Duration Noise)

Noise for phoneme pronunciation/duration. Default: 0.8

0 1.5

Global Speed Factor (Post-TTS Resample)

Overall speed change after TTS generation by resampling.

0.5 2

Examples (Piper TTS - Click to Run)

Input Text	Length Scale (Prosody/Speed)	Noise Scale (Variability)	Noise W (Phoneme Duration Noise)	Global Speed Factor (Post-TTS Resample)	Enable Gemini Text Enhancement	Gemini Enhancement Prompt	Google API Key (if not in secrets or startup config failed)