Sume TTS output_format: the mp3 44.1 kHz 128 kbps default and options

If you send no output_format, Sume TTS returns an mp3 at 44,100 Hz and 128 kbps. You can change the container to wav or raw, pick a sample rate from six values, and for mp3 pick a bit rate from five. The schema is strict, so any other field in output_format is rejected.

Everything the schema allows

Each container takes its own fields. Mp3 takes sample_rate and bit_rate. Wav takes sample_rate and an encoding that defaults to pcm_s16le. Raw has no defaults: you must send both sample_rate and encoding.

Picking one

For a video voiceover, the default mp3 is usually enough, and at 128 kbps a minute is about 960 KB. A 32 kbps mp3 is about 240 KB a minute, which can work for speech-only drafts. Choose wav when you will process the audio further and do not want another lossy encode. Choose raw with pcm_mulaw or pcm_alaw at 8000 Hz when the destination is a phone system, and remember a raw file has no header, so your player needs to be told the rate and encoding.

• Default for finished video: leave output_format out.
• Editing in a DAW: wav.
• Phone prompts: raw, pcm_mulaw, 8000 Hz.
• Draft listening on a slow link: mp3 at 32000 or 64000 bps.

Sizing a batch

File size is easy to estimate for mp3 because the bit rate is fixed. At 128 kbps a ten-minute narration is about 9.6 MB. At 192 kbps it is about 14.4 MB. At 64 kbps it is about 4.8 MB. Multiply by your episode count to see what your storage and delivery bill will look like, and remember do not assume the audio URL from a job is a permanent archive, so copy finished files to storage you control.

One caution

Output format does not change the price, which is by character. It does change file size and what your downstream tools accept, so test one file in your own player and editor before a batch.