Azure AI Speech text-to-speech vs Resemble AI Text-to-Speech API

Azure AI Speech text-to-speech

Real-time synthesis keeps neither the input text nor the output audio. An Azure subscription needs a card, even for the free F0 tier.

Resemble AI Text-to-Speech API

OpenAPI file in JSON and YAML, llms.txt and Markdown pages. Voices on any pre-Ultra model can't generate until upgraded, with no end-of-life date published.

Azure AI Speech text-to-speech

1. Send SSML with <speak> and <voice>, and set X-Microsoft-OutputFormat and User-Agent.
2. On 429 retry with backoff, and try the voice's home region or another region rather than asking for more quota.
3. Keep each real-time request under 10 minutes of audio, or use batch synthesis.
4. Use Entra ID tokens instead of resource keys where the agent runs inside Azure.
5. Cache the voice list per region, since it returns hundreds of entries at once.

Resemble AI Text-to-Speech API

1. Check the voice's model before synthesis, since voices on pre-Ultra models fail until upgraded.
2. Keep each synchronous request under 2,000 characters.
3. Decode audio_content from base64 on /synthesize, or call /stream for raw WAV chunks.
4. Send Authorization: Bearer, since some doc examples leave out the prefix.
5. Log the request ID with every failure, since the error body has no code.