Say

The Say command allows you to play text-to-speech (TTS) messages to a caller. It converts text to speech and then renders it in a voice back to the caller. Say is useful in cases where it's difficult to pre-record a prompt for any reason. When translating text to speech, the Say action will make assumptions about how to pronounce numbers, dates, times, amounts of money and other abbreviations.

FreeClimb now supports** enhanced voice engines** for more natural and customizable speech, while still keeping the classic version available for compatibility.

Say can only be done on a Call when not in Conference.

Note: Say command always flushes the DTMF buffer.

Versions of `Say`

You now have three ways to use the Say command:

Classic (default) – Simple, quick TTS with basic controls.

freeclimb.standard Engine– Uses Microsoft UCMA voices for more control (voices, cultures, and content types).

freeclimb.neural Engine– Powered by Coqui for advanced, lifelike TTS with SSML effects (pitch, rate, volume, etc.).

ElevenLabs Engine – AI-powered voices with highly natural, expressive speech and multilingual support.

Quick Decision Guide

Choose **Classic **if you need fast, lightweight TTS with minimal setup.
Choose `freeclimb.standard if you want Microsoft UCMA voices, accents, and cultural variations.
Choose `freeclimb.neural if you need lifelike, expressive speech and full SSML control.
Choose ElevenLabs if you want the most natural, human-like voices with emotional range and multilingual options.

Feature / Attribute	Classic	freeclimb.standard	freeclimb.neural (NEW 🚀)	ElevenLabs(In Development)
Engine name	(none)	`freeclimb.standard`	`freeclimb.neural`	ElevenLabs
Powered By	Built-in FreeClimb TTS	Microsoft UCMA	Coqui Neural TTS	ElevenLabs
text	✅ Limit 4KB	✅ Limit 4KB	✅ Limit 256 Characters	4KB
language	✅ (`en-US` default, must match table)`	❌(cannot be used when engine provided)	Set in `engine.parameters.language`(default `en-US`)
Voice selection	Implied from `language`	✅ Choose from 25_ named voices (`Voice`)	✅ Choose voice (default : `Eve`)
Culture/Locale	Controlled by `language`	✅ Explicitly set via `Culture`	✅ Controlled by `parameters.language`
Content type	Plain text only	`text/plain` or `application/ssml+xml`	`text` or `ssml`
SSML support	❌	✅ https://www.w3.org/TR/speech-synthesis/	✅	https://elevenlabs.io/docs/api-reference/text-to-speech/stream
Prosody controls	❌	Limited (UCMA SSML only)	✅ Full Coqui SSML support
privacyMode	✅ Hides text in logs	✅ Hides text in logs	✅ Hides text in logs
Best for...	Quick & simple TTS	Regional voices and UCMA compatibility	Natural, expressive speech with SSML effects

Nesting Rules

Say can exist as a standalone command or as a nested command. It does not allow barge-in unless nested within the following commands: GetDigits, GetSpeech. The file will always be played to completion unless nested or if it is nested and not interrupted by barge-In.

Classic `Say`

Attribute	Description
text (string, required)	The message to be played to the caller using TTS. The size of the string is limited to 4 KB (or 4,096 bytes). An empty string will cause the command to be skipped.
loop (integer, optional, default = 1)	Number of times the `text` is said. Specifying '0' causes the Say action to loop indefinitely. If there is a Call hangup or Barge-in, a loop will end even if it has not achieved the number of times initially specified.
language (string, optional, default = en-US)	Language and (by implication) the locale to use. This implies the accent and pronunciations to be used for the TTS. The complete list of valid values for the language attribute is shown below. Note: Language codes are case-sensitive and must be specified as shown in the table.
privacyMode (boolean, optional, default = false)	Indicates if the request contains sensitive information which should be hidden. When set to `true`, the contents of the `text` attribute will be replaced with the string "xxxxx" in the logs. It's important to note that `privacyMode` is set at the command level, meaning it will not inherit any value from a parent command.

freeclimb.standard Engine

Description: Uses Microsoft UCMA voices for more natural speech.

Attribute	Description
text (string, required)	The message to be played to the caller using TTS. The size of the string is limited to 4 KB (or 4,096 bytes). An empty string will cause the command to be skipped.
loop (integer, optional, default = 1)	How many times the text repeats. 0 = loop forever until hangup/barge-in.
language (string, optional, default = en-US)	name:`freeclimb.standard` parameters: Voice (default = ZiraPro) Culture(default = en-US) Content-Type(`text/plain` or `application/ssml+xml`)
privacyMode (boolean, optional, default = false)	Indicates if the request contains sensitive information which should be hidden. When set to `true`, the contents of the `text` attribute will be replaced with the string "xxxxx" in the logs. It's important to note that `privacyMode` is set at the command level, meaning it will not inherit any value from a parent command.

Voice/Culture Options

Voice	Culture
Herena	ca-ES
Helle	da-DK
Hedda	de-DE
Hayley	en-AU
Heather	en-CA
Hazel	en-GB
Heera	en-IN
Helen	en-US
ZiraPro	en-US
Helena	es-ES
Hilda	es-MX
Heidi	fi-FI
Harmonie	fr-CA
Hortense	fr-FR
Lucia	it-IT
Harkua	ja-JP
Heami	ko-KR
Hulda	nb-NO
Hanna	nl-NL
Paulina	pl-PL
Helosia	pt-BR
Helia	pt-PT
Elena	ru-RU
Hedvig	sv-SE
HuiHui	zh-CN
HunYee	zh-HK
HanHan	zh-TW

Note: Language codes are case-sensitive and must be specified as shown in the table.

Examples

Example (plain text):

[
  {
    "Say": {
      "text": "Welcome to FreeClimb.",
      "engine": {
        "name": "freeclimb.standard",
        "parameters": {
          "Voice": "Helen",
          "Culture": "en-US",
          "Content-Type": "text/plain"
        }
      }
    }
  }
]

Example (SSML):

[
  {
    "Say": {
      "text": "<prosody rate='50%'>Welcome to Freeclimb!</prosody>",
      "engine": {
        "name": "freeclimb.standard",
        "parameters": {
          "Voice": "Helen",
          "Culture": "en-US",
          "Content-Type": "application/ssml+xml"
        }
      }
    }
  }
]

freeclimb.neural Engine (NEW 🚀)

Description: This engine delivers lifelike voice and advanced speech control.

Attribute	Description
text (string, required)	The message to be played to the caller using TTS. The size of the string is limited to 4 KB (or 4,096 bytes). An empty string will cause the command to be skipped.
loop (integer, optional, default = 1)	How many times the `text` repeats. `0` = loop forever until hangup/barge-in.
language (string, optional, default = en-US)	name:`freeclimb.neural` parameters: Voice (default = Eve) textType(`text` or `ssml`) language(default = en-US)
privacyMode (boolean, optional, default = false)	Indicates if the request contains sensitive information which should be hidden. When set to `true`, the contents of the `text` attribute will be replaced with the string "xxxxx" in the logs. It's important to note that `privacyMode` is set at the command level, meaning it will not inherit any value from a parent command.

Neural Features:

SSML Support with <say-as>
Prosody controls for:
- Volume (increase/decrease)
- Speaking rate (faster/slower)
- Pitch adjustments
- Contour shaping (change pitch gradually over time)

Example (plain text):

[
  {
    "Say": {
      "text": "Welcome to FreeClimb.",
      "engine": {
        "name": "freeclimb.neural",
        "parameters": {
          "voice": "Eve"
        }
      }
    }
  }
]

Example (SSML with prosody):

[
  {
    "Say": {
      "text": "<prosody rate='50%'>Welcome to Freeclimb!</prosody>",
      "engine": {
        "name": "freeclimb.neural",
        "parameters": {
          "voice": "Eve",
          "textType": "ssml"
        }
      }
    }
  }
]

ElevenLabs Engine (In Development)

Description: This engine text into natural, human-like speech for narration, voiceovers, and conversations.

Attribute	Description
text (string, required)	The message to be played to the caller using TTS. The size of the string is limited to 4 KB (or 4,096 bytes). An empty string will cause the command to be skipped.
loop (integer, optional, default = 1)	How many times the `text` repeats. `0` = loop forever until hangup/barge-in.
language (string, optional, default = en-US)	name:`freeclimb.neural` parameters: (https://elevenlabs.io/docs/api-reference/text-to-speech/stream )
privacyMode (boolean, optional, default = false)	Indicates if the request contains sensitive information which should be hidden. When set to `true`, the contents of the `text` attribute will be replaced with the string "xxxxx" in the logs. It's important to note that `privacyMode` is set at the command level, meaning it will not inherit any value from a parent command.

Example (plain text)

{
  "Say": {
    "text": "Hello World",
    "engine": {
      "name": "ElevenLabs",
      "parameters": {
        "apply_text_normalization": false,
        "language_code": "en",
        "voice_id": "o11yegU3CL24TZ1qcm6b"
      }
    }
  }
}

Versions of Say