Convert SHN to SPX

Why Convert SHN to SPX?

Converting SHN to Speex serves a specialized purpose: extracting and compressing spoken content from concert recordings at extremely low bitrates. Speex is a speech-optimized codec, not a general music format, so this conversion is appropriate primarily for stage banter, audience announcements, spoken introductions, and other voice-centric segments from live recordings. For these specific use cases, Speex produces remarkably compact files with clear vocal reproduction.

Concert recordings often contain valuable spoken content beyond the music itself. Band members introduce songs, make dedications, interact with the audience, and share stories between musical passages. Historians and researchers studying live music culture may want to extract and archive these spoken segments efficiently. Speex at 16-24 kbps produces intelligible, clear speech in files that are 95% smaller than the original lossless SHN data.

Speex also remains relevant for legacy VoIP and embedded systems that specifically require the Speex codec. Some older conferencing platforms, Asterisk PBX installations, and industrial voice recording systems accept only Speex input. If you need to integrate concert audio into one of these legacy systems — for example, creating a phone-based concert information line or embedding audio in an older VoIP application — SHN-to-Speex conversion is necessary.

It is important to understand that Speex will significantly degrade the musical content of concert recordings. The CELP codec is modeled on the human vocal tract and does not faithfully reproduce instruments, harmonics, or the spatial ambience of a live venue. For general music listening, use FLAC, MP3, AAC, OGG, or Opus instead. Reserve Speex conversion for cases where speech intelligibility is the goal and music fidelity is not required.

Key Benefits of Converting SHN to SPX:

• Ultra-Compact: Speech at 8-24 kbps — files are 95%+ smaller than SHN
• Speech Optimized: CELP coding models human voice for clear vocal reproduction
• Noise Suppression: Built-in processing reduces background noise from recordings
• Open Source: Patent-free, royalty-free, part of Xiph.Org ecosystem
• Legacy VoIP: Compatible with Asterisk, FreeSWITCH, and older conferencing
• Low CPU: Decodes on minimal hardware including embedded systems
• Voice Detection: Built-in VAD compresses silence efficiently

Practical Examples

Example 1: Stage Banter Archive

Scenario: A researcher studying Jerry Garcia's between-song conversations needs to extract and archive spoken segments from hundreds of SHN concert recordings in the most compact format possible.

Source: Extracted banter clips from GD SHN recordings
Conversion: SHN → SPX (wideband, 24 kbps)
Result: Hours of speech in minimal storage

Banter archive:
  Original SHN clips: 15 GB (500 banter segments)
  Speex output: 450 MB (same 500 segments)
  → 97% reduction while preserving speech clarity
  + Voice activity detection skips applause gaps
  + Searchable by show date and venue in Ogg tags
  + Jerry's voice clearly intelligible at 24 kbps

Example 2: Concert Information Phone System

Scenario: A venue operates a legacy Asterisk-based phone system that plays back concert information and audio clips to callers, requiring Speex format for its VoIP codec.

Source: SHN concert audio clips and announcements
Conversion: SHN → SPX (narrowband, 16 kbps)
Result: Phone-quality audio for Asterisk PBX

VoIP integration:
+ Speex is native to Asterisk conferencing system
+ 8 kHz narrowband matches telephone audio quality
+ 16 kbps fits within VoIP bandwidth allocation
+ Callers hear concert clips and show information
+ Minimal server storage for audio prompts library

Example 3: Spoken Word Podcast from Concert Archives

Scenario: A podcast producer creates episodes about live music history using spoken segments extracted from SHN recordings, needing lightweight audio for supplementary web content.

Source: Selected spoken clips from SHN concert archives
Conversion: SHN → SPX (ultra-wideband, 44 kbps)
Result: Web-optimized speech clips

Podcast supplement use:
+ Ultra-compact files for web page embedding
+ 44 kbps UWB provides good speech clarity
+ Spoken introductions and band commentary preserved
+ Minimal bandwidth for mobile web visitors
+ Used alongside higher-quality music clips in other formats

Frequently Asked Questions (FAQ)

Q: What is Speex and why is it speech-specific?

A: Speex uses Code-Excited Linear Prediction (CELP), a technique that models audio as if it were produced by the human vocal tract. This makes it extremely efficient for speech — capturing voice at very low bitrates with good intelligibility. However, this same approach means it poorly represents instruments, harmonics, and complex audio scenes. Speex was created by Jean-Marc Valin (who later co-created Opus) specifically for voice communication, not music.

Q: Will my concert music sound good in Speex?

A: No. Speex significantly degrades musical content because its CELP model is optimized for voice frequencies and characteristics. Instruments will sound muffled, distorted, or artificial. Audience ambience and room acoustics will be lost. Use Speex only for spoken content (stage banter, announcements, spoken introductions). For music from concert recordings, use FLAC, MP3, AAC, OGG Vorbis, or Opus instead.

Q: Is Speex still relevant or should I use Opus?

A: The Xiph.Org Foundation officially recommends Opus over Speex for all new applications. Opus was designed as Speex's successor and outperforms it in every dimension — speech quality, music quality, latency, and bitrate efficiency. The only reasons to use Speex are compatibility with legacy systems that specifically require it (older VoIP platforms, embedded devices with Speex-only decoders) or when working with existing Speex infrastructure.

Q: What sample rate and bitrate should I use?

A: For speech extraction from concerts, wideband (16 kHz, 24 kbps) provides the best balance of quality and size — voice sounds natural and clear. Ultra-wideband (32 kHz, 44 kbps) adds minimal quality improvement for speech. Narrowband (8 kHz, 16 kbps) sounds like a telephone call but produces the smallest files. Choose based on your quality requirements and storage constraints.

Q: Can I play SPX files on my computer or phone?

A: VLC plays Speex files on all platforms — Windows, macOS, Linux, Android, and iOS. foobar2000 and mpv also support Speex decoding. However, native operating system media players generally do not support .spx files. Web browser support is also absent. If playback convenience matters, consider Opus instead, which has much broader native support while still being efficient for speech content.

Q: How does Speex noise suppression help with concert recordings?

A: Speex includes built-in preprocessing features like noise suppression and automatic gain control. When converting spoken segments from concert recordings, these features can help reduce background crowd noise, venue reverb, and instrument bleed, making the speech clearer. However, this processing also removes ambient concert atmosphere. For pure speech clarity from noisy recordings, this is a benefit; for preserving the concert experience, it is a limitation.

Q: How small are Speex files compared to other formats?

A: Extremely small. At wideband 24 kbps, one minute of speech takes about 180 KB in Speex. Compare this to: SHN (roughly 10 MB/min), FLAC (roughly 5 MB/min), MP3 320k (roughly 2.4 MB/min), Opus 128k (roughly 960 KB/min). Speex produces files 55 times smaller than SHN for speech content. This extreme compression is the main reason to choose Speex — when storage and bandwidth are severely constrained.

Q: How long does SHN to Speex conversion take?

A: Very fast — Speex encoding is lightweight by design, originally intended for real-time voice communication. A 70-minute recording converts in 1-3 minutes, roughly 30-50 times faster than real-time. The process involves SHN decoding (fast), resampling to Speex's supported rates (trivial), and Speex encoding (very fast). Batch processing of many segments completes quickly.