Paxa Labs

Paxa Labs: Speech and language API for Thai, English and Mandarin

Paxa Labs offers text to speech, speech to text, OCR and translation through one API, with models built for Thai and for speech that moves between Thai, English and Mandarin. Its flagship model, Paxa TTS Flash v2, speaks all three languages in every voice, even within a single sentence, and takes delivery directions written into the text.

Key Features

• Multilingual text to speech: Paxa TTS Flash v2 offers 50 voices, and each one speaks English, Thai and Mandarin, including mixed-language lines.
• Directed delivery: Tags written into the text, such as [whispering], [laughs] or [apologetic, sincere], set how a line is spoken.
• Streaming: Audio arrives as chunks over plain HTTP, or over a live websocket that starts speaking while an LLM is still writing its reply.
• OpenAI-compatible endpoint: The speech endpoint answers the OpenAI SDK at /v1/audio/speech, so switching only needs a new base URL.
• Word timestamps: Speech and transcription can return word, sentence or utterance spans for captions and read-along interfaces.
• Speech to text (research preview): Paxa STT Lite transcribes recordings of up to 120 minutes in Thai, English, Mandarin and other languages, with subtitles in SRT or VTT, speaker turns and a live transcription mode.
• Spoken or written Thai: One option writes numbers, dates and times in Thai transcripts either as words, as they were said, or as digits, as Thai is written.
• OCR and extraction: Reads receipts, tax invoices, forms and reports in Thai, English, Chinese and other languages into Markdown, structured data or the fields a schema names.
• AI translation: Translates between Thai, English, Chinese and other languages, with formality settings, glossaries and document context.

Use Cases

• Call centers and voice agents: Thai and multilingual voices for phone support, with streaming for live conversations.
• Audiobooks, podcasts and e-learning: Long-form narration and lessons in Thai, English or Mandarin.
• Captions and call recordings: Transcripts, subtitles and speaker turns for meetings, interviews, lectures and video.
• Document processing: Extracting data from Thai invoices, contracts, ID forms and archives.

Pricing

One credit balance covers every API. A free tier gives 100 credits with no card required. Monthly plans are Starter ($6), Standard ($21) and Pro ($99), and pay-as-you-go refills cost $10 per 10,000 credits. Speech costs 10 credits per 1,000 characters, transcription 8.33 credits per minute, OCR 6.5 credits per page and translation 25 credits per 1,000 characters. Enterprise plans are available on request.

Getting Started

• Website: paxalabs.com
• Documentation: paxalabs.com/docs
• Pricing: paxalabs.com/pricing

Paxa Labs suits teams that build voice and document products for Thai users and need speech, transcription, OCR and translation from a single API.