Speechmatics

Broad language coverage; Melia-1 code-switches across 56 languages.

All prices are the vendor’s public list price — 0% markup. Extra provider knobs pass through untouched via provider_params.

Model	Modes	List price	Diarization	Word timings
speechmatics/enhanced	streaming · batch	$1.04/hr	✓	✓
speechmatics/standard	streaming · batch	$0.7992/hr	✓	✓
speechmatics/melia-1	batch	$0.129/hr	✓	✓

Provider options

Reach past the unified surface with provider_params — forwarded streaming → transcription_config fields; batch → transcription_config fields. Typed in the SDKs as {provider}Params interfaces (providerParams option / provider_params= kwarg).

Param	Type	Default	Applies to	What it does
max_delay	number	4	streaming	Max seconds before a final is emitted
max_delay_mode	flexible · fixed	—	streaming	Whether max_delay may stretch to finish entities
speaker_diarization_config	object	—	streaming · batch	{max_speakers, speaker_sensitivity, prefer_current_speaker}
additional_vocab	array	—	streaming · batch	[{content, sounds_like[]}] — up to 1000 entries with pronunciations
enable_entities	boolean	true	streaming · batch	Emit typed entities (numbers, dates) in results
punctuation_overrides	object	—	streaming · batch	Tune permitted marks and sensitivity
conversation_config	object	—	streaming	{end_of_utterance_silence_trigger: 0–2s} — must be < max_delay

Try it

curl -s https://api.speechrouter.ai/v1/audio/transcriptions \
  -H "Authorization: Bearer $SPEECHROUTER_API_KEY" \
  -F model=speechmatics/enhanced \
  -F file=@audio.wav

_{Generated from the gateway catalog — the billing engine’s own source of truth.}